Skip to main content
Internet Archive's 25th Anniversary Logo

Full text of "Heterosis : a record of researches directed toward explaining and utilizing the vigor of hybrids"

See other formats


Heterosis 


HETEROSIS 


A  record  of  researches  directed  toward  explaining 
and  utilizing  the  vigor  of  hybrids 


Edited  by 

JOHN  W.   GOWEN 

Professor  of  Genetics 
Iowa  Stafe  College 


I  tc  l» 


IOWA  STATE  COLLEGE  PRESS 

AMES   •   IOWA 


Copyrlghi  1952  by  The  Iowa  Sfate  College  Press 

All  rights  reserved.   Composed  and  printed  in  fhe 

United  States  of  America 


Preface 


Heterosis  grew  out  of  a  desire  on  the  part  of  Iowa  State  College  to  gather  to- 
gether research  workers  from  marginal  fields  of  science,  each  with  something 
to  contribute  to  a  discussion  of  a  central  problem  of  major  national  interest. 
The  problem  of  heterosis,  as  synonymous  in  large  part  with  that  of  hybrid 
vigor,  formed  a  natural  theme  for  discussion.  As  the  reader  will  note,  many 
fields  of  science  have  contributed  or  stand  to  make  significant  contributions 
to  the  subject.  Major  steps  in  the  advance  have  led  to  divergent  views  which 
may  be  rectified  only  through  joint  discussions  followed  by  further  research. 
The  conference  of  students  of  this  problem  was  held  June  15  to  July  20, 1950. 

In  furnishing  the  opportunity  for  these  discussions  by  active  research 
workers  in  the  field,  Iowa  State  College  hoped:  to  facilitate  summarization 
and  clarification  of  the  accumulated  data  on  the  subject,  to  encourage  formu- 
lation and  interpretation  of  the  observations  in  the  light  of  present  day  bio- 
logical information,  to  stimulate  further  advances  in  the  controlled  success- 
ful utilization  and  understanding  of  the  biological  processes  behind  the  phe- 
nomenon of  heterosis,  and  to  increase  the  service  rendered  by  this  discovery 
in  expanding  world  food  supply. 

Iowa  has  a  direct,  vested  interest  in  heterosis.  Today  the  agricultural 
economy  of  the  state  is  based  upon  hybrid  corn.  The  scene  portraying  a  hy- 
bridization block  of  corn,  shown  here,  is  familiar  to  all  who  travel  within  the 
state  as  well  as  to  those  in  surrounding  regions,  for  this  method  of  corn 
breeding  has  been  shown  to  be  surprisingly  adaptable  and  useful  in  producing 
more  food  per  acre  over  wide  areas  of  the  world's  agricultural  lands. 

Iowa's  indebtedness  to  heterosis,  generated  through  crossing  selected  and 
repeatedly  tested  inbred  strains,  is  well  known.  Few  outside  the  workers  in  the 
field  realize  the  full  magnitude  of  this  debt. 

With  the  progressive  introduction  of  hybrid  corn  in  1936  there  came  a 
steady  increase  in  corn  yields  over  both  the  former  yields  and  over  the 
yields  of  other  agricultural  crops,  as  that  of  tame  hay,  which  were  not  sub- 
ject to  this  genetic  method  of  yield  improvement.  It  seems  likely  that  in  no 
other  period  of  like  years  has  there  been  such  an  increase  in  food  produced 
over  so  many  acres  of  land.  The  return  from  hybrid  corn  has  been  phenome- 
nal, but  it  is  now  evidently  approaching  an  asymptotic  value.  It  behooves  us 
to  find  out  as  much  as  possible  about  the  techniques  and  methods  which 


VI 


PREFACE 


made  these  advances  possible.  Even  more  we  should  determine  what  is  going 
on  within  the  breeding  and  physiological  systems  through  which  heterosis 
finds  expression,  if  further  increases  in  yields  are  to  be  obtained  or  better 
systems  of  breeding  are  to  be  developed. 

Toward  this  end  the  conference  topics  were  arranged  under  four  major 


j^'u'X  ''■*Je.Ai 


mi 


'.»*.. 


Controlled  heterosis  in  the  making  through  pollinations  and  fertilizations  of  selectively 
purified  genetic  strains  of  corn  (maize).  (From  G.  F.  Sprague.) 


lA 

z      a: 


E  z 
o 


O  V>  -J 

Z   -10- 


1 1, 


4  O  <t 


K  -  Z 

u  >-  e 


>  *  . 


100 

/ 

'^' 

^ 

90- 

/ 

/ 

PERCENTAGE     OF  CORN   ACREAGE 
PLANTED    WITH    HYBRID    SEED 

80 

/ 

/ 

70 

/ 

' 

1 

60 

1 

i 

1 

/ 

\ 

-. 

s. 

/ 

\ 

/ 

\ 

50 

f 

-7] 

/ 

ACRE     YIELD 
OF     CORN 

r\ 

/ 

\ 

/ 

\ 

^ 

40 
30 
20 

/' 

\ 

1 

\ 

/ 

w 

/ 

N. 

\ 

1 

\ 

/ 

I 

\ 

/ 
/ 
/ 

--' 

■  -- 

^-. 

_.\ 

/ 

/ 

.-' 

\ 

\ 
\ 

> 

/ 

/ 
/ 

\ 

1 
1 
1 

\ 

— ""  " 

"'^  ^. 

1 

ACRE      YIELD     OF 
TAME     HAY / 

/^- 

y' 

\ 

1 

10 

/ 

/ 

/ 

0 



_^     — 

—   -^ 

1930 


1935 


1940 
YEARS 


1945 


1950 


Trends  in  acre  production  of  maize  before  and  after  heterosis  was  in  use  and  tame  hay 
over  which  there  has  been  no  such  breeding  control  1930-1950.  (From  G.  F.  Sprague.) 


PREFACE  vii 

headings.  The  early  history  and  development  of  the  heterosis  concepts  and 
the  cy  to  logical  aspects  of  the  problem  occupied  the  first  week.  The  contribu- 
tions of  physiology,  evolution,  and  specific  gene  or  cytoplasmic  effects  to  the 
vigor  observed  in  hybrids  were  dealt  with  the  second  week.  The  third  week's 
meetings  covered  postulated  gene  interactions,  as  dominance,  recombination, 
and  other  possible  gene  effects.  During  the  fourth  week  breeding  systems  and 
methods  of  utilizing  and  evaluating  heterosis  effects  were  considered.  In  the 
final  week  the  students  considered  the  problems  that  lie  ahead  and  recent 
methods  of  meeting  them. 

At  each  daily  conference  the  speaker  of  the  day  presented  a  formal  morn- 
ing lecture  covering  his  subject.  In  the  afternoon,  he  led  a  conference  session 
on  the  subject  of  the  morning  lecture.  At  this  time,  all  present  had  an  oppor- 
tunity to  participate. 

Accompanying,  and  as  a  supplement  to  the  Heterosis  Conference,  a 
Methods  Workshop  was  held  from  July  3  to  July  13.  The  Workshop  was  de- 
voted to  recent  techniques  for  evaluating  the  kinds  of  data  which  occur  fre- 
quently in  animal  breeding  experiments.  Workshop  meetings  were  organized 
by  Professor  R.  E.  Comstock  of  North  Carolina  State  College  and  Professor 
Jay  L.  Lush  of  Iowa  State  College. 

The  meetings  were  led  by  men  from  several  institutions  besides  Iowa 
State  College.  Professors  Oscar  Kempthorne,  Jay  L.  Lush,  C.  R.  Henderson, 
G.  E.  Dickerson,  L.  N.  Hazel,  F.  H.  Hull,  A.  E.  Bell,  A.  M.  Button,  J.  Bruce 
Griffing,  C.  C.  Cockerham,  F.  H.  W.  Morley,  R.  M.  Koch,  and  A.  L.  Rae 
contributed  much  to  this  phase  of  the  program.  It  is  with  regret  that  it  is 
impossible  to  present  the  meat  of  the  methods  presented  and  developed  in 
the  Workshop  and  the  afternoon  discussions.  To  many,  this  material  con- 
tributed much  to  the  merit  of  the  conference  and  the  use  to  which  the  results 
were  put  later. 

In  the  field  of  worth-while  living,  as  well  as  to  see  heterosis  in  operation, 
conferees  were  guests,  on  various  weekends,  of  three  nearby  companies  putting 
heterosis  to  the  practical  test  of  commercial  seed  stock  production  in  crops 
and  live  stock — the  Ames  Incross  Company,  the  Farmers  Hybrid  Corn 
Company,  and  the  Pioneer  Hi-Bred  Corn  Company. 

Finally,  the  organization  of  the  conference  was  the  product  of  the  joint 
effort  of  the  genetic  group  of  Iowa  State  College.  This  group  transcends  all 
departmental  lines  having  as  the  common  interest  what  goes  on  in  inher- 
itance. They  were  Jay  L.  Lush,  G.  F.  Sprague,  Oscar  Kempthorne,  S.  S. 
Chase,  Janice  Stadler,  L.  N.  Hazel,  A.  W.  Nordskog,  Iver  Johnson,  W.  A. 
Craft,  J.  Bruce  Griffing,  and  John  W.  Gowen. 

In  last  analysis  it  was  the  interest  of  the  audience  and  their  participations 
in  the  discussions  that  made  the  Conference  worth  while.  The  papers  cover- 
ing material  presented  by  the  leaders  of  these  discussions  follow. 


Table  of  Contents 


1.  Early  Ideas  on  Inbreeding  and  Crossbreeding   ....   conway  zirkle  1 

2.  Beginnings  of  the  Heterosis  Concept       .      .      .      George  Harrison  shull  14 

3.  Development  of  the  Heterosis  Concept H.  k.  hayes  49 

4.  Preferential  Segregation  in  Maize M.  M.  rhoades  66 

5.  Inbreeding  and  Crossbreeding  in  Seed  Development    .      .      .     r.  a.  brink  81 

6.  Physiology  of  Gene  Action  in  Hybrids w.  Gordon  whaley  98 

7.  Hybrid  Nutritional  Requirements william  j.  robbins  114 

8.  Origin  of  Corn  Belt  Maize  and  Its  Genetic  Significance 

EDGAR  ANDERSON  and  WILLIAM  L.   BROWN  124 

9.  Heterosis  in  Population  Genetics                  .  Adrian o  a.  buzzati-tra verso  149 

10.  Fi.xing  Transgressive  Vigor  in  Nicotiana  Rustica    .     .       harold  h.  smith  161 

11.  Hybridization  in  the  Evolution  of  Maize      .      .      .     paul  c.  mangelsdorf  175 

12.  Biochemical  Models  of  Heterosis  in  Neurospora      .      .    sterling  emerson  199 

13.  Nature  and  Origin  of  Heterosis th.  dobzhansky  218 

14.  Plasmagenes  and  Chromogenes  in  Heterosis       .      .      .        donald  f.  jones  224 

15.  Specificity  of  Gene  Effects m.  r.  irwin  236 

16.  Genetics  and  Cytology  of  Saccharomyces     ....   carl  c.  lindegren  256 

17.  Genetic  Implications  of  Mutations  in  S.  Typhiniurium      .      .  h.  h.  plough  267 

18.  Dominance  and  Overdominance james  f.  crow  282 

19.  Gene  Recombination  and  Heterosis leroy  powers  298 

20.  Gene  Interaction  in  Heterosis A.  j.  mangelsdorf  320 

21.  Inbred  Lines  for  Heterosis  Tests? Gordon  e.  dickerson  3vS0 

22.  Specific  and  General  Combining  Ability       ....       c.  R.  Henderson  352 

23.  Rotational  Crossbreeding  and  Heterosis L.  m.  winters  371 

24.  Gamete  Selection  for  Specific  Combining  Ability 

e.  l.  pinnell,  e.  h.  rinke,  and  h.  k.  hayes  378 

25.  Monoploids  in  Maize sherret  s.  chase  389 

26.  Early  Testing  and  Recurrent  Selection g.  F.  sprague  400 

27.  Heterosis  in  a  New  Population E.  j.  wellhausen  418 

28.  Recurrent  Selection  and  Overdominance fred  H.  hull  451 

29.  Hybrid  Vigor  in  Drosophila John  w.  gowen  474 

30.  Estimation  of  Average  Dominance  of  Genes 

R.  e.  comstock  and  h.  f.  robinson  494 

Bibliography 517 

Index 537 


IX 


67867 


CONWAY  ZIRKLE 

University  of  Pennsylvania 


OGIC^^ 


Chapter   1 

Early  Ideas  on  Inbreeding 
and  Crossbreeding 


In  tracing  the  historical  background  of  a  great  scientific  advance  or  dis- 
covery, the  historian  nearly  always  has  the  opportunity  of  showing  that  the 
scientists  who  receive  the  credit  for  the  work  are  really  late-comers  to  the 
field,  and  that  all  the  basic  principles  and  facts  were  known  much  earlier. 
Finding  these  earlier  records  is  always  something  of  a  pleasure;  comparable, 
perhaps,  to  the  pleasure  a  systematist  experiences  in  extending  the  range 
of  some  well  known  species. 

The  historian  may  be  tempted,  in  consequence,  to  emphasize  these  earlier 
contributions  a  little  too  strongly  and  to  re-assign  the  credits  for  the  scientific 
advances  which  have  been  made.  In  the  present  state  of  the  history  of  sci- 
ence, it  requires  only  a  little  searching  of  the  records  to  discover  contributions 
which  have  been  overlooked  and  which  are  very  pertinent  to  the  advance 
in  question.  This  wealth  of  data,  which  accumulates  almost  automatically, 
seems  to  deserve  emphasis.  But  great  steps  forward  generally  are  made 
not  by  the  discovery  of  new  facts,  important  as  they  are,  or  by  new  ideas, 
brilliant  as  they  may  be,  but  by  the  organization  of  existing  data  in  such 
a  way  that  hitherto  unperceived  relationships  are  revealed,  and  by  incor- 
porating the  pertinent  data  into  the  general  body  of  knowledge  so  that  new, 
basic  principles  emerge. 

For  example,  even  so  monumental  a  work  as  Darwin's  Origin  of  Species 
contains  few  facts,  observations  or  even  ideas  which  had  not  been  known 
for  a  long  time.  The  work  of  many  pre-Darwinians  now  appears  important, 
especially  after  Darwin's  synthesis  had  shown  its  significance.  Of  course, 
this  does  not  belittle  Darwin  in  the  slightest.  It  only  illustrates  the  way 
science  grows. 

The  emergence  of  the  scientific  basis  of  heterosis  or  hybrid  vigor  is  no 

1 


2  CONWAY  ZIRKLE 

exception.  Practically  all  of  its  factual  background  was  reported  before 
Mendel's  great  contribution  was  discovered.  Even  workable  methods  for 
utilizing  hybrid  vigor  in  crop  production  were  known,  but  it  was  not  until 
the  classic  post-Mendelian  investigations  of  Shull,  East,  and  Jones  were 
completed,  that  heterosis  took  its  proper  place  in  genetics.  The  following 
discussion  of  the  importance  of  heterosis  will  be  confined  to  its  pre-Men- 
delian  background. 

Heterosis  can  be  described  as  a  special  instance  of  the  general  principles 
involved  in  inbreeding  and  outbreeding.  To  fit  it  into  its  proper  niche,  we  will 
trace  first  the  evolution  of  our  ideas  on  the  effects  of  these  two  contrasting 
types  of  mating.  Since  our  earliest  breeding  records  seem  limited  to  those 
of  human  beings  and  primitive  deities,  we  will  start  with  the  breeding 
records  of  these  two  forms. 

Hybrid  vigor  has  been  recognized  in  a  great  many  plants  during  the 
last  two  hundred  years.  We  will  therefore  describe  briefly  what  was  known 
of  its  influence  on  these  plants.  Because  heterosis  has  reached  its  greatest 
development  in  Zea  mays,  we  will  trace  briefly  the  pre-Mendelian  genetics 
of  this  plant,  and  show  how  the  facts  were  discovered  which  have  been  of 
such  great  scientific  and  economic  importance. 

The  ill  effects  of  too-close  inbreeding  have  been  known  for  a  long  time. 
Indeed,  Charles  Darwin  (1868)  believed  that  natural  selection  had  pro- 
duced in  us  an  instinct  against  incest,  and  was  effective  in  developing  this 
instinct  because  of  the  greater  survival  value  of  the  more  vigorous  offspring 
of  exogamous  matings.  One  of  his  contemporaries,  Tylor  (1865),  noted  that 
many  savage  tribes  had  tabooed  the  marriage  of  near  relatives,  and  he 
assumed  that  they  had  done  so  because  they  had  noticed  the  ill  effects  of 
inbreeding.  The  Greeks  looked  upon  certain  marriages  between  near  rela- 
tives as  crimes.  This  has  been  known  almost  universally  ever  since  Freud 
popularized  the  tragedy  of  King  Oedipus.  At  present,  we  outlaw  close  in- 
breeding in  man,  and  our  custom  is  scientifically  sound. 

We  are  apt  to  be  mistaken,  however,  if  we  read  into  the  standards  of  our 
distant  preceptors  the  factual  knowledge  which  we  have  today.  The  in- 
tellectual ancestors  of  European  civilization  approved  of  inbreeding  and 
actually  practiced  it  on  supposedly  eugenic  grounds.  The  fact  that  their 
genetics  was  unsound  and  their  eugenic  notions  impractical  is  irrelevant. 
They  had  their  ideals,  they  were  conscientious  and  they  did  their  duties. 
The  Pharaohs  married  their  own  sisters  when  possible  so  that  their  god- 
like blood  would  not  be  diluted.  Marriage  between  half  brother  and  sister 
was  common  in  other  royal  families  of  the  period.  Actually,  as  we  shall  see, 
the  two  great  pillars  of  European  thought,  Hebrew  morality  and  Greek 
philosophy,  endorsed  inbreeding  as  a  matter-of-course. 

The  Hebrews,  who  derived  mankind  from  a  single  pair,  were  compelled 
to  assume  that  the  first  men  born  had  to  marry  their  sisters — as  there  were 


EARLY  IDEAS  ON   INBREEDING  AND  CROSSBREEDING  3 

then  no  other  women  on  the  earth.  Indeed  Adam  and  Eve  themselves  were  not 
entirely  unrelated.  The  marrying  of  a  sister  was  obviously  respectable,  and 
it  seems  to  have  occurred  routinely  among  the  Hebrews  and  their  ancestors 
for  several  thousand  years.  Abraham's  wife,  Sarah,  was  also  his  sister.  At 
times  even  closer  inbreeding  took  place.  Abraham's  nephew,  Lot,  impreg- 
nated his  own  two  daughters.  The  latter  instances  occurred,  however,  under 
exceptional  circumstances — and  Lot  was  drunk.  But  as  late  as  the  time  of 
King  David,  brother-sister  marriages  took  place.  The  imbroglio  between 
David's  children,  Tamar,  Ammon,  and  Absalom,  shows  that  a  legal  mar- 
riage between  half-brother  and  sister  would  then  have  been  a  routine  oc- 
currence. 

The  Greeks  also  could  hardly  have  had  scruples  against  inbreeding,  as 
evinced  by  the  pedigrees  they  invented  for  their  gods.  Their  theogony  shows 
many  instances  of  the  closest  inbreeding  possible  for  either  animals  or  gods 
in  which  the  sexes  are  separate.  Zeus,  the  great  father  of  the  gods,  married 
his  sister,  Hera.  Their  parents,  Kronos  and  Rhea,  also  were  brother  and 
sister,  and  were  in  turn  descended  from  Ouranos  and  Gaea,  again  brother 
and  sister.  Thus  the  legitimate  offspring  of  Zeus — Hebe,  Ares,  and  Hephaes- 
tus— were  the  products  of  three  generations  of  brother-sister  mating. 
Moreover,  the  pedigrees  of  the  Greek  heroes  show  an  amount  of  inbreeding 
comparable  to  that  in  our  modern  stud  books  for  race  horses.  They  were 
all  related  in  one  way  or  another  and  related  to  the  gods  in  many  ways.  A 
single  example  will  be  cited.  Zeus  was  the  father  of  Herakles  and  also  his 
great-great-grandfather  on  his  mother's  side.  Herakles'  great-great-grand- 
mother, Danae,  who  had  found  such  favor  in  the  eyes  of  Zeus,  was  herself 
descended  from  Zeus  through  two  different  lines.  With  immortals,  back- 
crossing  offered  no  real  problems. 

East  and  Jones  (1919)  have  pointed  out  that  close  inbreeding  was  com- 
mon among  the  Athenians  even  at  the  height  of  their  civilization.  These 
scientists  were  of  the  opinion  that  most  of  the  freemen  in  Attica  were 
rather  closely  related  to  each  other.  Marriage  between  half  brother  and 
sister  was  permitted,  and  marriage  between  uncle  and  niece  fairly  common. 
A  Grecian  heiress  was  nearly  always  taken  as  a  wife  by  one  of  her  kinsmen 
so  that  her  property  would  not  be  lost  to  the  family.  Common  as  inbreeding 
was  during  the  flowering  of  Greek  culture,  it  was  as  nothing  compared  with 
the  inbreeding  which  occurred  in  the  period  after  the  Trojan  War  and  before 
the  true  historical  period.  In  this  intervening  time,  Greece  was  divided  into 
innumerable  independent  political  units,  many  of  them  minute.  One  island 
six  miles  long  and  two  miles  wide  contained  three  separate  kingdoms. 
Political  boundaries  as  well  as  bays,  mountains,  and  seas  were  functional, 
isolating  mechanisms;  and  the  Greeks  were  separated  into  many  small 
breeding  units  for  fifteen  to  twenty  generations.  Isolation  was  never  com- 
plete, however,  and  there  were  enough  wandering  heroes  to  supply  some 


4  CONWAY  ZIRKLE 

genie  migration.  There  were  also  some  mass  migrations  and  amalgamations 
of  different  tribes.  The  general  situation  was  startlingly  close  to  the  condi- 
tions which  Sewall  Wright  (1931)  describes  as  the  optimum  for  rapid 
evolution. 

We  may  be  tempted  to  explain  as  cause  and  effect  what  may  be  only  an 
accidental  relationship  in  time;  and,  while  recognizing  that  it  is  far  fetched, 
to  ascribe  the  sudden  appearance  of  what  Galton  called  the  ablest  race  in 
history  to  the  ideal  conditions  for  evolution  which  their  ancestors  had.  We 
would  also  like  to  consider,  as  the  necessary  preliminary  to  the  hybrid  vigor, 
that  period  of  inbreeding  which  preceded  the  flowering  of  Grecian  culture. 
This  hybrid  vigor  we  would  like  to  recognize  as  an  important  factor  in  the 
production  of  the  great  geniuses  who  flourished  in  the  later,  larger  city 
states  of  Greece. 

So  much  for  the  classical  attitude  toward  endogamy.  It  slowly  changed, 
and  exogamy  which  had  always  existed  became  the  exclusive  custom.  At 
the  time  of  Sophocles,  all  forms  of  inbreeding  were  not  considered  ethical 
and  pleasing  to  the  gods.  The  sin  of  Oedipus  lay  in  his  having  made  a  for- 
bidden backcross  rather  than  in  mere  inbreeding  which  was  lawful.  We  do 
not  find  any  records  of  degeneracy  appearing  in  his  children — indeed  his 
daughter  Antigone  was  a  model  of  feminine  virtue.  It  seems  that  close 
human  inbreeding  came  to  an  end  without  its  ill  effects  ever  having  been 
recognized. 

The  Nordics  also  were  unaware  of  any  degeneracy  inherent  in  inbreeding. 
Their  great  god  Wotan  included  a  bit  of  inbreeding  in  his  plan  for  creating 
a  fearless  hero  who  could  save  even  the  gods  themselves  from  their  im- 
pending fate.  Wotan  started  the  chain  reaction  by  begetting  Siegmund  and 
Sieglinde,  twin  brother  and  sister.  The  twins  were  separated  in  infancy. 
They  met  again  as  adults  and,  recognizing  their  relationship,  had  an  il- 
legitimate affair — begetting  the  hero  Siegfried.  Although  Siegfried  was  not 
exactly  an  intellectual  type,  he  was  certainly  not  a  degenerate — represent- 
ing rather  the  ideal  male  of  a  somewhat  primitive  culture. 

As  the  centuries  passed,  incest  was  extended  to  cover  brother-sister 
mating,  even  when  the  parties  involved  were  unaware  of  their  relationship. 
There  is  no  need  to  cite  here  the  many  examples  of  the  later  tragedies  based 
upon  this  plot.  It  soon  became  an  almost  universally  accepted  standard  in 
literature,  from  epics  to  novels.  The  luckless  Finnish  hero,  Kullervo  {The 
Kalevala,  Rune  XXXV),  thus  brought  disaster  to  his  family  by  seducing  his 
sister  unknowingly.  Defoe's  long  suffering  heroine  Moll  Flanders  (1722) 
had  to  abandon  an  apparently  successful  marriage  when  she  discovered  that 
her  husband  was  her  brother.  On  the  other  hand,  as  late  as  1819,  Lord 
Byron  defended  brother-sister  marriage  passionately  in  his  drama  Cain — 
but  this  was  a  scandalous  exceptioii  to  the  rule.  The  marriage  of  kin  nearer 


EARLY  IDEAS  ON  INBREEDING  AND  CROSSBREEDING  5 

than  first  cousins  had  become  legally  and  morally  taboo.  Perhaps  we  may 
follow  Westermarck  in  assuming  that  endogamy  became  passe,  not  because 
its  biological  ill  effects  were  recognized,  but  because  men  knew  their  kins- 
women too  well  to  marry  one  of  them  if  they  could  possibly  get  a  wife 
elsewhere. 

It  is  possible  that  we  have  thus  far  paid  too  much  attention  to  inbreeding 
and  outbreeding  in  man.  Our  excuse  is  that  there  are  almost  no  other  records 
of  inbreeding  from  classical  times.  There  are  no  plant  records,  of  course,  for 
sex  in  plants  was  not  understood  in  spite  of  the  general  practices  of  caprifica- 
tion  and  hand  pollination  of  the  date  palm.  Records  of  inbreeding  and  out- 
crossing in  domestic  animals  are  almost  completely  lacking  even  in  the 
copious  agricultural  literature  of  the  Romans.  Aristotle's  History  of  Animals. 
576al5  (Thompson  1910)  does  state  that  horses  will  cover  both  their  mothers 
and  their  daughters  ".  .  .  and,  indeed,  a  troup  of  horses  is  only  considered 
perfect  when  such  promiscuity  of  intercourse  occurs" — but  he  seems  to 
be  almost  alone  in  referring  to  the  subject.  Later  on  in  the  same  book 
(630b30)  he  cited  a  happening  which  we  quote. 

The  male  camel  declines  intercourse  with  its  mother;  if  his  keeper  tries  compulsion,  he 
evinces  disinclination.  On  one  occasion,  when  intercourse  was  being  declined  by  the 
young  male,  the  keeper  covered  over  the  mother  and  put  the  young  male  to  her;  but,  when 
after  the  intercourse  the  wrapping  had  been  removed,  though  the  operation  was  completed 
and  could  not  be  revoked,  still  by  and  by  he  bit  his  keeper  to  death.  A  story  goes  that  the 
king  of  Scythia  had  a  highly-bred  mare,  and  that  all  her  foals  were  splendid;  that  wishing 
to  mate  the  best  of  the  young  males  with  the  mother,  he  had  him  brought  to  the  stall  for 
the  purpose;  that  the  young  horse  declined;  that,  after  the  mother's  head  had  been  con- 
cealed in  a  wrapper  he,  in  ignorance,  had  intercourse;  and  that,  when  immediately  after- 
wards the  wrapper  was  removed  and  the  head  of  the  mare  was  rendered  visible,  the  young 
horse  ran  away  and  hurled  himself  down  a  precipice. 

This  behavior  of  the  stallion  was  considered  so  remarkable  that  it  was 
described  by  Aelian,  Antigonus,  Heirocles,  Oppian,  Pliny,  and  Varro. 
Varro  confused  the  tradition  and  made  the  horse  bite  his  keeper  to  death. 

It  is  fairly  safe  for  us  to  assume  that  in  both  classical  and  medieval  times 
the  flocks  and  herds  were  greatly  inbred.  Transportation  difficulties  would 
have  insured  inbreeding  unless  its  evil  effects  were  realized,  and  we  have  at 
least  negative  evidence  that  they  were  not.  Varro,  who  gave  many  detailed 
directions  for  the  breeding  of  all  domestic  animals,  does  not  even  mention 
the  question  of  kinship  between  sire  and  dam.  We  do  have  an  interesting 
literary  allusion  by  Ovid,  however,  to  the  routine  inbreeding  of  domestic 
animals  in  his  account  of  the  incest  of  Myrrha  in  the  tenth  book  of  the  Meta- 
morphoses. The  affair  between  Myrrha  and  her  father  Cinyras  was  like  that 
of  Oedipus  and  his  mother  Jocasta.  The  fates  had  decreed  that  Myrrha 
should  become  the  mistress  of  her  father.  Torn  by  her  unholy  desires  she 
debates  the  matter  with  her  conscience.  Her  better  nature  argues  (From 
the  metrical  translation  of  Brookes  More,  1922): 


6  CONWAY  ZIRKLE 

But  what  more  could  be  asked  for,  by  the  most 
Depraved?  Think  of  the  many  sacred  ties 
And  loved  names,  you  are  dragging  to  the  mire; 
The  rival  of  your  mother,  will  you  be 
The  mistress  of  your  father,  and  be  named 
The  sister  of  your  son,  and  make  yourself 
The  mother  of  your  brother? 

In  Stating  the  other  side  of  the  case  Myrrha  describes  the  "natural"  in- 
breeding of  animals. 

A  crime  so  great — If  it  indeed  is  crime. 

I  am  not  sure  it  is — I  have  not  heard 

That  any  God  or  written  law  condemns 

The  union  of  a  parent  and  his  child. 

All  animals  will  mate  as  they  desire — 

A  heifer  may  endure  her  sire,  and  who 

Condemns  it?  And  the  happy  stud  is  not 

Refused  by  his  mare-daughters:  the  he-goat 

Consorts  unthought-of  with  the  flock  of  which 

He  is  the  father;  and  the  birds  conceive 

Of  those  from  whom  they  were  themselves  begot. 

Happy  are  they  who  have  such  privilege! 

Malignant  men  have  given  spiteful  laws; 

And  what  is  right  to  Nature  is  decreed 

Unnatural,  by  jealous  laws  of  men. 

But  it  is  said  there  are  some  tribes  today. 
In  which  the  mother  marries  her  own  son; 
The  daughter  takes  her  father;  and  by  this, 
The  love  kind  nature  gives  them  is  increased 
Into  a  double  bond. — Ah  wretched  me! 

The  debate  ends  as  we  would  expect,  and  in  due  course  Myrrha  is  de- 
livered of  an  infant  boy  who  certainly  showed  none  of  the  ill  effects  of  the  in- 
breeding which  produced  him.  He  grew  up  to  be  quite  an  Adonis.  In  fact 
he  was  Adonis. 

We  can  profitably  skip  to  the  late  eighteenth  century  before  we  pursue 
further  the  matter  of  inbreeding.  This  was  the  period  when  Bakewell  was 
emphasizing  the  importance  of  breeding  in  improving  farm  animals,  when 
the  various  purebreds  were  beginning  to  emerge,  and  when  the  efficacy  of 
artificial  selection  was  beginning  to  be  understood. 

By  the  beginning  of  the  nineteenth  century,  practical  attempts  to  im- 
prove the  different  breeds  of  cattle  led  to  intensive  inbreeding.  A  prize  bull 
would  be  bred  to  his  own  daughters  and  granddaughters.  At  first,  the  breed- 
ers seemed  to  believe  that  a  selection  of  the  very  best  individuals  followed 
by  intensive  inbreeding  was  the  quickest  method  for  improving  the  stock. 
On  theoretical  grounds  this  seemed  to  be  the  case,  and  great  advances 
were  actually  made  by  this  method — but  sooner  or  later  something  always 


EARLY   IDEAS  ON   INBREEDING  AND  CROSSBREEDING  7 

happened.  The  inbred  stock  seemed  to  grow  sterile,  but  vigor  could  be  re- 
established by  outcrossing.  The  actual  cause  of  degeneracy  in  the  inbreds 
was  not  understood  until  Mendelian  inheritance  was  discovered,  but  the 
remedial  procedures  of  the  practical  breeders  could  hardly  have  been  im- 
proved on.  We  owe  to  them  the  basis  of  our  finest  stocks.  They  inbred  to 
add  up  and  concentrate  desirable  qualities  and  then  crossbred  to  prevent 
degeneration,  then  inbred  again  and  crossed  again,  all  the  time  selecting 
their  breeding  stocks  most  carefully.  Charles  Darwin  (1868)  described  this 
process  most  accurately  and  listed  the  pertinent  publications. 

There  was  a  striking  divergence  in  this  work  between  theory  and  prac- 
tice, which  is  just  as  well,  as  the  only  theories  available  at  the  time  were  in- 
adequate. Those  breeders  who  held  that  inbreeding  was  the  suninmm  bonum 
did  not  hesitate  to  crossbreed  when  the  occasion  demanded,  and  those  who 
emphasized  the  virtues  of  hybridization  inbred  whenever  inbreeding  gave 
them  the  opportunity  of  adding  up  desirable  qualities.  Darwin,  himself, 
stated,  "Although  free  crossing  is  a  danger  on  the  one  side  which  everyone 
can  see,  too  close  inbreeding  is  a  hidden  danger  on  the  other."  We  await 
the  twentieth  century  for  a  real  improvement  in  breeding  methods. 

The  first  plant  hybrid  was  described  as  such  in  1716,  and  during  the  next 
forty-five  years  many  descriptions  of  hybrid  plants  were  published.  Some 
attempts  were  even  made  to  produce  new  varieties,  but  in  retrospect  the 
work  seems  somewhat  dilettante. 

From  1761  to  1766,  Josef  Gottlieb  Koelreuter  (1766)  published  the  several 
parts  of  his  well-known  classic,  and  plant  hybridization  was  put  upon  a 
different  and  more  scientific  basis.  His  investigation  of  hybridization  was 
intensive,  systematic,  and  scientific.  He  described,  among  other  things, 
hybrid  vigor  in  interspecific  crosses  in  Nicotiana,  Dianthus,  Verbascum, 
Mirabilis,  Datura,  and  other  genera  (East  and  Jones,  1919).  He  also  observed 
floral  mechanisms  which  insured  cross  pollination  and  assumed  in  conse- 
quence that  nature  had  designed  plants  to  benefit  from  crossbreeding.  It  is 
worth  emphasizing  that  hybrid  vigor  in  plants  was  first  described  by  the 
person  who  first  investigated  plant  hybrids  in  detail.  Koelreuter  continued 
to  publish  papers  on  plant  hybrids  until  the  early  nineteenth  century. 

Meanwhile  other  contributions  had  been  made  to  our  knowledge  of  the 
effects  of  outcrossing  and  the  mechanism  for  securing  it.  In  1793,  Sprengel 
depicted  the  structure  of  flowers  in  great  and  accurate  detail,  and  showed 
how  self  pollination  was  generally  avoided.  In  1799,  Thomas  Andrew  Knight 
described  hybrid  vigor  as  a  normal  consequence  of  crossing  varieties  and 
developed  from  this  his  principle  of  anti-inbreeding.  Other  hybridizers 
noted  the  exceptional  vigor  of  many  of  their  creations.  Indeed,  hybrid 
vigor  in  plants  was  becoming  a  commonplace.  Among  the  botanists  who 
recorded  this  vigor  were:  Mauz  (1825),  Sageret  (1826),  BerthoUet  (1827), 
Wiegmann  (1828),  Herbert  (1837),  and  Lecoq  (1845).  Gartner  (1849)  was 


8  CONWAY  ZIRKLE 

especially  struck  by  the  vegetative  luxuriance,  root  development,  height, 
number  of  flowers  and  hardiness  of  many  of  his  hybrids. 

Naudin  (1865)  found  hybrid  vigor  in  twenty-four  species  crosses  out  of  the 
thirty-five  which  he  made  within  eleven  genera.  In  Datura  his  results  were 
spectacular.  In  reciprocal  crosses  between  D.  Stramonium  and  D.  Tatula 
the  offspring  were  twice  the  height  of  the  parents.  Knowledge  of  plant 
hybridization  was  increasing  more  rapidly  at  this  time  than  the  biologists 
knew,  for  this  was  the  year  in  which  Mendel's  (1865)  paper  Versuche  iiber 
Pflanzen-Hyhriden  appeared.  Mendel  discovered  hybrid  vigor  in  his  pea 
hybrids  and  described  it  as  follows: 

The  longer  of  the  two  parental  stems  is  usually  exceeded  by  the  hybrid,  a  fact  which  is 
possibly  only  attributable  to  the  greater  luxuriance  which  appears  in  all  parts  of  the 
plants  when  stems  of  very  different  lengths  are  crossed.  Thus,  for  instance,  in  repeated 
experiments,  stems  of  1  ft.  and  6  ft.  in  length  yielded  without  exception  hybrids  which 
varied  in  length  between  6  ft.  and  1\  ft. 

We  shall  cite  but  one  more  scientist  who  wrote  on  the  general  subject  of 
hybrid  vigor  in  plants.  This  is  Charles  Darwin,  whose  Cross  and  Self  Fertiliza- 
tion in  the  Vegetable  Kingdom  appeared  in  1876.  This  was  a  book  of  great 
importance  and  influence,  but  no  attempt  will  be  made  here  to  summarize 
this  work  of  nearly  five  hundred  pages.  At  the  beginning  of  his  concluding 
chapter,  Darwin  stated: 

The  first  and  most  important  conclusion  which  may  be  drawn  from  the  observations 
given  in  this  volume,  is  that  cross-fertilization  is  generally  beneficial  and  self-fertihzation 
injurious. 

There  is  a  special  reason  why  this  book  of  Darwin's  is  of  such  great 
importance  for  any  historical  background  to  heterosis.  Darwin  worked 
carefully  and  quantitatively  with  many  genera,  including  Zea  mays.  He 
measured  accurately  the  amount  of  hybrid  vigor  he  could  induce,  and  he  pub- 
lished his  data  in  full.  His  work  stands  in  the  direct  ancestral  line  to  the 
twentieth  century  research  on  the  subject,  and  the  great  advances  made 
from  1908  to  1919  are  based  solidly  on  this  work.  There  are  no  great  gaps 
in  the  steady  progress  and  no  gaps  in  the  literature. 

Zea  mays  was  brought  to  Europe  in  1493  by  Columbus  on  his  home- 
ward voyage.  This  was  sometime  before  the  great  herbals  were  written, 
so  our  first  descriptions  of  the  new  grain  are  to  be  found  in  the  books  of  the 
travelers  and  explorers.  Later,  Indian  corn  appeared  under  various  names 
in  the  early  herbals,  and  it  was  described  in  detail  in  the  famous  Krautehuch 
of  Tabernaemontanus,  first  published  in  1588.  The  author  obviously  yielded 
to  his  enthusiasm  in  devoting  five  and  a  half  folio  pages  to  corn  and  includ- 
ing thirteen  illustrations  in  his  treatment.  He  was  the  first  to  describe  the 
results  of  xenia — the  occurrence  of  difi'erent  colored  grains  on  the  same  ear — 
but  his  explanation  of  the  phenomenon  has  nothing  to  do  with  cross  pollina- 
tion. He  ascribed  it  directly  to  God  Almighty. 


EARLY  IDEAS  ON   INBREEDING  AND  CROSSBREEDING  9 

And  one  sees  an  especially  great  and  wonderful  mystery  in  these  spikes,  Gott  der  Ilerr, 
through  the  medium  of  nature  which  must  serve  everyone,  disports  himself  and  performs 
wonders  in  his  works  and  so  notably  in  the  case  of  this  plant  that  we  must  rightly  be 
amazed  and  should  learn  to  know  the  One  True  Eternal  God  even  from  his  creatures  alone. 
For  some  of  the  spikes  of  this  plant,  together  with  their  fruit,  are  quite  white,  brown  and 
blue  intermixed.  Thus,  some  rows  are  half  white,  a  second  series  brown  and  the  third  blue; 
and  some  grains,  accordingly  are  mixed  with  each  other  and  transposed.  Again,  sometimes 
one,  two,  or  three  rows  are  white,  the  next  rows  blue,  then  again  white  and  after  that 
chestnut-brown;  that  is,  they  are  interchanged  on  one  row  and  run  straight  through  on 
another.  Some  spikes  and  their  grains  are  entirely  yellow,  others  entirely  brown,  some  are 
white,  brown,  and  blue,  others  violet,  white,  black,  and  brown:  of  these  the  white  and 
blue  are  prettily  sprinkled  with  small  dots,  as  if  they  had  been  artistically  colored  in  this 
way  by  a  painter.  Some  are  red,  black,  and  brown,  with  sometimes  one  color  next  to  the 
other,  while  at  other  times  two,  three,  even  four  colors,  more  or  less,  are  found  one  next 
to  another  in  this  way. 

During  the  next  century  and  a  half,  many  other  descriptions  of  the 
occurrence  of  different  colored  grains  on  a  single  ear  were  published.  I  have 
found  about  forty  of  them  and  there  are  doubtless  many  more.  The  earliest 
correct  interpretation  of  this  phenomenon  had  to  await  the  eighteenth  cen- 
tury and  is  contained  in  a  letter  written  by  Cotton  Mather  in  1716.  Here 
the  different  colored  grains  occurring  together  on  an  ear  are  ascribed  to  a 
wind-born  intermixture  of  varieties.  This  letter  is  the  first  record  we  have  of 
plant  hybridization,  and  antedates  Fairchild's  description  of  a  Dianthus 
hybrid  by  one  year.  In  1724,  Paul  Dudley  also  described  hybridization  in 
maize,  and  he  was  able  to  eliminate  one  of  the  hypotheses  which  had  been 
used  to  explain  the  mixture.  As  a  broad  ditch  of  water  lay  between  the  mix- 
ing varieties,  he  could  show  that  the  mixed  colors  were  not  due  to  the  root- 
lets of  different  strains  fusing  underground,  a  view  held  at  the  time  by 
many  New  Englanders,  both  white  and  red. 

Hybridization  in  maize  was  described  again  in  1745  by  Benjamin  Cooke, 
in  1750  by  the  great  Swedish  traveler  and  naturalist,  Pehr  Kalm,  and  in 
1751  by  William  Douglass.  By  the  early  nineteenth  century,  knowledge  of 
plant  hybrids  was  widespread.  Plant  hybridization  was  becoming  a  routine 
practice,  and  there  is  little  doubt  that  different  varieties  of  maize  were 
crossed  many  times  by  American  farmers  who  did  not  record  their  breeding 
experiments  in  writing. 

Brown  and  Anderson  (1947,  1948)  have  recently  shown  that  the  modern 
races  now  grown  in  the  corn  belt  are  derived  from  both  the  northern  flint 
and  the  southern  dent  varieties.  Hybridization  in  corn  was  easy  to  perform 
and  the  results  were  easy  to  recognize.  The  intermixtures  of  colors  were  so 
spectacular  that  they  were  frequently  described,  by  Gallesio  (1806),  Burger 
(1808),  Sageret  (1826),  Gartner  (1827),  and  others. 

We  detour  briefly  here  into  some  of  the  technical  aspects  of  xenia.  Double 
fertilization  and  the  mixed  nature  of  the  endosperm  were  discovered  by  Na- 
waschin  in  1899.  In  1881,  Focke  introduced  the  term  xenia  but  he  used  it 
to  include  what  we  now  call  melaxenia.  Focke  collected  from  the  literature 


10  CONWAY  ZIRKLE 

many  supposed  instances  where  the  pollen  influenced  directly  the  color 
and  form  of  the  flowers,  the  flavor  and  shape  of  the  fruits,  and  the  color 
and  content  of  the  seeds.  How  many  of  these  cases  were  really  due  to  Men- 
delian  segregation  we  will  probably  never  know,  since  the  investigators  did 
not  know  enough  to  take  proper  precautions. 

We  can,  however,  divide  the  history  of  true  xenia  into  three  periods: 
first,  when  its  visible  effect  was  considered  a  lusus  naturae  (1588);  second, 
when  it  was  known  to  be  caused  by  foreign  pollen  (1716);  and  third,  when 
the  embryo  and  endosperm  were  recognized  as  two  different  structures  and 
when  the  influence  of  the  pollen  upon  the  latter  was  recorded  specifically. 
In  the  paragraph  on  Zea  in  the  section  on  xenia,  Focke  cites  the  work  of 
Vilmorin  (1867),  Hildebrand  (1868),  and  Kornicke  (1876),  who  described 
the  effect  of  pollen  on  the  endosperm. 

We  should  note  a  brief  comment  on  the  subject  which  has  been  overlooked 
and  is  earlier  than  the  papers  cited  by  Focke.  In  1858,  Asa  Gray  described 
xenia  in  maize.  He  reported  starchy  grains  in  ears  of  sweet  corn  and  many 
different  kinds  and  colors  of  grains  on  the  same  ear.  He  had  two  explana- 
tions for  this  occurrence:  (1)  cross  pollination  of  the  previous  year  and  (2) 
direct  action  of  the  pollen  on  the  ovules  of  the  present  year.  It  is  obvious 
that  by  ovules  he  did  not  mean  embryos.  This  may  be  the  earliest  authentic 
recognition  of  the  real  problem  of  xenia. 

In  reviewing  the  nineteenth  century  records  of  hybrid  vigor  in  Zea  mays, 
we  start  with  those  of  Charles  Darwin  (1876).  Darwin  planned  his  experi- 
ments most  carefully.  He  crossed  and  selfed  plants  from  the  same  stock,  and 
raised  fifteen  plants  from  each  of  the  two  types  of  seed  he  had  obtained. 
He  planted  the  seed  from  both  the  selfed  and  crossed  plants  in  the  same 
pots,  from  six  to  ten  plants  per  pot.  When  the  plants  were  between  one  and 
two  feet  in  height,  he  measured  them  and  found  that  the  average  height  of 
the  plants  from  the  selfed  seed  was  17.57  inches,  while  that  from  the  crossed 
seed  was  20.19  inches  or  a  ratio  of  81  to  100.  When  mature,  the  two  lots 
averaged  61.59  inches  and  66.51  inches,  respectively,  a  ratio  of  93  to  100. 
In  another  experiment  when  the  corn  was  planted  in  the  ground,  the  ratio 
of  the  selfed  to  the  crossed  was  80  to  100.  Darwin  called  in  his  cousin,  Francis 
Galton,  to  check  his  results  and  Galton  judged  them  to  be  very  good  after 
he  had  studied  the  curves  that  he  drew. 

The  direct  connection  between  Darwin's  work  and  our  present  hybrid 
corn  is  shown  by  Darwin's  influence  on  W^  J.  Beal  who  was  the  real  leader 
in  the  American  research  designed  to  improve  maize.  Beal  reviewed  Dar- 
win's book  in  1878,  and  even  wrote  an  article  which  was  little  more  than  a 
paraphrase  of  what  Darwin  had  published.  Beal's  own  contributions  ap- 
peared a  little  later. 

In  1880,  Beal  described  how  he  had  increased  the  yield  of  corn  on  a  large 
scale.  Two  stocks  of  the  same  type  of  corn  which  had  been  grown  a  hundred 


EARLY  IDEAS  ON   INBREEDING  AND  CROSSBREEDING  11 

miles  apart  for  a  number  of  years  were  planted  together  in  alternate  rows. 
All  of  one  stock  grown  in  this  field  was  detasseled  and  thus  it  could  not  be 
self  fertilized  but  could  produce  only  hybrid  seed.  The  tasseled  stalks  of  the 
other  lot  would  still  be  pure  bred  as  there  was  no  foreign  pollen  to  contami- 
nate their  ears  and  they  could  again  serve  as  a  parent  to  a  hybrid.  A  small 
amount  of  the  first  parental  stock  which  furnished  the  detasseled  stalks  was 
grown  apart  for  future  hybridization.  The  hybrid  seed  was  planted,  and 
produced  the  main  crop.  Beal  increased  his  yield  by  this  method  by  as 
much  as  151  exceeds  100.  This  method  and  these  results,  it  should  be 
emphasized,  were  published  in  1880. 

E.  Lewis  Sturtevant,  the  first  director  of  the  New  York  Agricultural 
Experiment  Station,  made  a  number  of  studies  of  corn  hybrids  starting  in 
1882.  His  findings  are  interesting  and  important  but  not  directly  applicable 
to  heterosis.  Singleton  (1935)  has  called  attention  to  this  work  and  to  the 
excellent  genetic  research  which  the  western  corn  breeders  were  carrying  on 
at  this  time — such  geneticists  as  W.  A.  Kellerman,  W.  T.  Swingle,  and 
Willet  M.  Hays.  They  anticipated  many  of  Mendel's  findings  and  described 
dominance,  the  reappearance  of  recessives  (atavisms),  and  even  Mendelian 
ratios  such  as  1  to  1  and  3  to  1.  They  were  all  concerned  with  practical 
results.  Hays  (1889),  in  particular,  tried  to  synthesize  superior  breeds  of 
corn  by  hybridizing  controlled  varieties. 

Sanborn  (1890)  confirmed  Beal's  results  and  reported  that  his  own 
hybrid  corn  yielded  in  the  ratio  of  131  to  100  for  his  inbred.  He  also  fol- 
lowed Real's  method  of  planting  his  parental  stocks  in  alternate  rows  and  of 
detasseling  one  of  them.  He  made  an  additional  observation  which  we  know 
now  is  important: 

It  is  this  outcrossed  seed  which  will  give  the  great  crops  for  the  next  year.  It  will  be 
noted  that  I  gained  twelve  bushels  i)er  acre  by  using  crossed  seed.  The  operation  is  simple 
and  almost  costless  and  will  pay  one  hundred  fold  for  the  cost  involved.  The  cross  must  be 
made  every  year  using  nen'  seed,  the  product  of  the  outcross  of  two  pure  seed.  (Italics  C.  Z.) 

If  our  farmers  had  known  of  this  discovery  reported  in  1890  they  might 
not  have  tried  to  use  their  own  hybrid  corn  as  seed. 

Singleton  (1941)  also  called  attention  to  a  pre-Mendelian  interpretation 
of  hybrid  vigor  by  Johnson  (1891)  which,  in  the  light  of  our  present  knowl- 
edge, deserves  more  than  passing  notice.  We  can  state  it  in  Johnson's  own 

words : 

That  crossing  commonly  gives  better  offspring  than  in-and-in  breeding  is  due  to  the 
fact  that  in  the  latter  both  parents  are  likely  to  possess  by  inheritance  the  same  imperfec- 
tions which  are  thus  intensified  in  the  progeny,  while  in  cross  breeding  the  parents  more 
usually  have  different  imperfections,  which  often,  more  or  less,  compensate  each  other  in 
the  immediate  descendants. 

We  come  next  to  a  j)ublication  of  G.  W.  McClure  (1892).  This  paper  is 
deservedly  famous,  and  its  many  contributions  are  incorporated  into  our 
modern  genetics  literature.  Here  we  shall  cite  only  the  observations  which 
pertain  to  heterosis.  McClure  noted  (1)  that  sterility  and  deformity  often 


12  CONWAY  ZIRKLE 

follow  selfing,  (2)  that  crossing  imparts  vigor,  (3)  that  it  is  impossible 
to  tell  in  advance  what  varieties  will  produce  corn  of  increased  size  when 
crossed,  (4)  that  what  appears  to  be  the  best  ear  does  not  always  produce 
the  largest  crops,  and  (5)  nearly  all  of  the  hybrid  corn  grown  a  second  year  is 
smaller  than  that  grown  the  first  year,  though  most  of  it  is  yet  larger  than 
the  average  size  of  the  parent  varieties. 

McClure  also  called  attention  to  the  fact  that  our  fine  varieties  of  fruits 
have  to  be  propagated  vegetatively,  and  hinted  that  the  deteriorations  of 
the  seedlings  from  fruit  trees  was  not  unrelated  to  a  like  deterioration  which 
occurred  in  the  seedlings  grown  from  hybrid  corn. 

The  year  following  McClure's  publication.  Morrow  and  Gardiner  (1893) 
recorded  some  very  pertinent  facts  they  had  discovered  as  a  result  of  their 
field  experiments  with  corn.  They  reported  that,  "In  every  instance  the 
yield  from  the  cross  is  greater  than  the  average  from  the  parent  varieties: 
the  average  increase  per  acre  from  the  five  crosses  [they  had  made]  being 
nine  and  a  half  bushels."  They  noted  further  in  a  paper  published  later  the 
same  year  that,  "It  seems  that  cross  bred  corn  gives  larger  yields  at  least 
for  the  first  and  second  years  after  crossing  than  an  average  of  the  parent 
varieties,  but  how  long  this  greater  fruitfulness  will  last  is  undetermined." 
Gardiner  continued  the  work  and  in  1895  published  the  data  he  obtained 
by  repeating  the  experiments.  He  found  that  in  four  of  six  cases  the  yield 
was  greater  in  the  cross,  the  average  being  twelve  bushels  per  acre. 

We  now  come  to  the  great  corn  breeding  research  project  which  was 
undertaken  at  the  University  of  Illinois  in  1895  by  Eugene  Davenport 
and  P.  G.  Holden.  Both  of  these  scientists  had  been  students  of  Beal  and 
were  interested  in  his  work  on  inbreeding  and  cross  breeding  maize.  We 
are  indebted  to  Professor  Holden  for  an  account  of  this  work  which  he  printed 
privately  in  1948.  This  account  gives  us  valuable  historic  data  not  to  be 
found  elsewhere,  as  most  of  the  University  of  Illinois  records  were  destroyed 
by  fire. 

An  intensive  series  of  inbreeding  experiments  was  undertaken  by  Holden, 
and  later  on  the  inbred  lines  were  crossed.  Hybrid  vigor  was  noted,  and  it 
was  found  in  addition  that  the  crosses  between  different  inbred  lines  differed 
widely  in  their  yield  and  in  their  general  desirability.  The  main  purpose  of 
the  experiments  was  to  find  out  how  to  use  controlled  crossing  early  and 
effectively.  After  Holden  left  Illinois  in  1900,  the  project  was  taken  over  by 
C.  G.  Hopkins,  a  chemist,  who  was  interested  in  increasing  the  protein  con- 
tent of  maize.  He  hired  as  his  assistant  in  1900  a  young  chemist  named 
Edward  Murray  East,  whom  we  shall  hear  about  later. 

Our  account  of  the  background  of  heterosis  is  coming  to  an  end  as  the 
beginning  of  the  twentieth  century  makes  a  logical  stopping  point.  We  should 
mention,  however,  the  great  hybrid  vigor  discovered  by  Webber  (1900) 
when  he  crossed  a  Peruvian  corn,  Cuzco,  with  a  native  variety,  Hickory 


EARLY  IDEAS  ON  INBREEDING  AND  CROSSBREEDING  13 

King.  The  average  height  of  the  parental  stocks  was  8  feet  3  inches  while 
the  cross  averaged  12  feet  4  inches,  an  increase  of  4  feet  1  inch. 

The  next  year  Webber  (1901)  called  attention  to  the  marked  loss  of  vigor 
in  corn  from  inbreeding.  From  100  stalks  of  selfed  corn  he  obtained  46 
ears  weighing  9.33  pounds,  while  from  100  stalks  obtained  from  crossing 
different  seedlings  he  obtained  82  ears  weighing  27.5  pounds.  When  he 
attempted  to  "fix"  his  Cuzco-Hickory  King  hybrid  by  selling  he  got  a  great 
loss  of  vigor  and  almost  complete  sterility,  but  when  he  crossed  the  different 
seedlings  there  was  little  loss  of  vigor.  He  concluded  that  to  fix  hybrids 
one  should  not  self  the  plants. 

In  1900,  the  discovery  of  Mendel's  long-forgotten  paper  was  announced. 
Both  Hugo  de  Vries  and  C.  Correns,  two  of  the  three  discoverers  of  Mendel, 
published  papers  on  Zea  mays  and  all  future  work  on  Indian  Corn  was  on  a 
somewhat  different  level. 

SUMMARY  OF  KNOWLEDGE  OF  HYBRID  VIGOR  AT 
BEGINNING  OF  20th  CENTURY 

1.  Inbreeding  reduces  vigor  and  produces  many  defective  and  sterile  indi- 
viduals which  automatically  discard  themselves. 

2.  Cross  breeding  greatly  increases  vigor  both  in  interspecific  and  inter-   < 
varietal  hybrids.  Crossing  two  inbred  stocks  restores  the  lost  vigor  and 
frequently  produces  more  vigor  than  the  stocks  had  originally. 

3.  All  inbred  stocks  do  not  produce  the  same  amount  of  vigor  when  crossed.  ^ 
Certain  crosses  are  far  more  effective  than  others. 

4.  The  simplest  method  of  hybridizing  Zea  on  a  large  scale  is  to  plant  two 
stocks  in  alternate  rows  and  to  detassel  one  stock.  The  hybrid  corn  grown 
from  the  detasseled  stock  produces  the  great  yields. 

5.  Hybridization  must  be  secured  each  generation  if  the  yield  is  to  be  kept 
up,  although  a  second  generation  of  open  pollinated  corn  may  still  be 
better  than  the  original  parental  stocks. 

6.  In  inbreeding,  both  parents  are  apt  to  have  the  same  defects  which  are  V 
intensified  in  the  offspring.  The  cause  of  hybrid  vigor  is  that  in  crosses 
the  parents  usually  have  different  defects  which  tend  to  compensate  for 
each  other  in  the  immediate  progeny. 

7.  The  fact  that  hybrid  vigor  in  Zea  is  not  permanent  but  decreases  if  the 
hybrids  are  open-pollinated,  seems  to  be  related  to  the  fact  that  fruit 
trees,  whose  desirable  qualities  are  preserved  by  vegetative  propagation, 
produce  seedlings  which  are  inferior. 


GEORGE  HARRISON  SHULL 

Princefon   University 


Chapter  2 

Beginnings  of 
the  Heterosis  Concept 


The  heterosis  concept  was  first  definitely  recognized  in  the  work  with  hybrid 
corn.  Before  attempting  to  define  this  concept,  however,  we  will  take  a  brief 
look  at  some  of  the  observations  of  early  workers  which  indicated  the  prob- 
able presence  of  heterosis,  and  where  recognition  of  heterosis  as  an  important 
biological  principle  might  have  been  expected. 

The  first  hybridizer  of  plants,  Dr.  J.  G.  Koelreuter,  noted  some  impres- 
sive examples  of  excessive  luxuriance  in  his  Nicotiana  hybrids.  These  were 
isolated  observations  which  suggested  no  theory  as  to  why  these  hybrids 
should  exceed  their  parents  in  size  and  general  vigor.  Koelreuter  cannot  be 
said  to  have  had  a  heterosis  concept.  Probably  every  conscious  producer  of 
hybrids  since  Koelreuter's  time  has  made  similar  observations  of  the  exces- 
sive vigor  of  some  hybrids  over  their  parents,  so  that  such  hybrid  vigor  has 
ceased  to  cause  surprise.  But  the  general  acceptance  of  hybrid  vigor  as  a  nor- 
mal phenomenon  did  not  estabUsh  a  heterosis  concept.  It  was  merely  the 
summational  effect  of  oft-repeated  experience. 

Thomas  Andrew  Knight  noted  the  deterioration  of  some  of  the  old  stand- 
ard horticultural  varieties,  and  concluded  that  such  varieties  have  a  natural 
life-span  and  gradually  decline  as  the  result  of  advancing  senility.  He  saw 
that  such  decline  makes  it  necessary  to  develop  new  varieties  which  will  start 
off  with  the  vigor  of  youth.  Although  Knight  himself  produced  many  such 
new  varieties,  some  of  which  were  produced  by  hybridization,  it  is  not  ap- 
parent that  he  thought  of  hybridization  as  an  agency  for  the  production  of 
such  new  vigor.  Although  he  advanced  a  theory  concerning  physiological 
vigor  and  its  decline,  he  did  not  recognize  the  heterosis  concept. 

Luther  Burbank  also  produced  numerous  varieties,  often  following  inten- 

14 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  15 

tional  hybridizations,  and  it  is  easy  to  recognize  heterosis  as  a  potent  factor 
in  the  remarkable  values  displayed  by  many  of  these  new  varieties.  But 
while  Burbank  made  great  use  of  hybridizations  in  his  plant  breeding  work, 
he  did  not  recognize  hybridization,  as  such,  as  the  source  of  the  large  size 
and  remarkable  vigor  of  his  new  varieties.  For  him  the  role  of  hybridization, 
aside  from  the  bringing  together  of  desirable  qualities  possessed  separately 
by  the  two  chosen  parents,  was  merely  the  "breaking  of  the  types."  In  this 
way  the  variability  in  subsequent  generations  was  greatly  increased,  thus 
enlarging  the  range  of  forms  from  among  which  to  select  the  most  desirable 
for  recognition  as  New  Creations. 

There  are  many  other  important  observations  and  philosophical  considera- 
tions that  bear  a  close  relationship  to  our  current  understanding  of  heterosis, 
and  which  antedated  the  recognition  of  heterosis.  It  would  take  us  too  far 
afield,  however,  to  discuss  these  related  observations  at  length.  We  can 
make  only  this  passing  reference  to  the  highly  significant  work  of  Charles 
Darwin  in  demonstrating  that  cross-fertilization  results,  in  many  cases,  in 
increased  size,  vigor,  and  productiveness  as  compared  with  self-fertilization 
or  with  other  close  inbreeding  within  the  same  species. 

Darwin  did  not  recognize  this  increased  vigor  as  identical  with  hybrid 
vigor,  nor  specifically  attribute  it  to  the  differences  between  the  uniting 
gametes.  To  him  it  only  demonstrated  a  method  which  would  inevitably 
preserve  by  natural  selection  any  variation  that  might  occur — whether  me- 
chanical or  physiological — which  would  make  cross-fertilization  more  likely 
or  even  an  obligate  method  of  reproduction.  With  heterosis  established  as  a 
recognized  pattern  of  behavior,  or  type  of  explanation,  we  can  now  interpret 
Darwin's  demonstrated  superiority  of  crossbreds  as  examples  of  the  occur- 
rence of  heterosis.  We  may  go  even  further  and  include  the  whole  field  of 
sexual  reproduction  in  showing  the  advantages  of  heterosis.  These  result 
from  the  union  of  two  cells — the  egg  and  the  sperm — extremely  difTerentiated 
physiologically,  and  in  all  dioecious  organisms  also  dififerentiated  genetically. 

Let  us  briefly  consider  several  investigations  which  foreshadowed  the 
procedures  now  used  in  growing  hybrid  corn — for  somewhere  in  the  course 
of  this  work  with  corn  the  heterosis  principle  was  first  definitely  recognized. 

Two  techniques  are  characteristically  associated  with  the  work  of  the 
"hybrid-corn  makers."  Uncritical  commentators  have  mistakenly  considered 
these  techniques  synonymous  with  the  development  of  the  hybrid-corn  pro- 
gram itself.  These  are  (a)  cross-pollination  by  interplanting  two  different 
lines  or  varieties,  and  the  detasseling  of  one  of  these  lines  which  then  sup- 
plies the  seed  to  be  planted;  and  {b)  controlled  self-pollination. 

In  deciding  what  part  these  two  methods  played  in  the  develoi)mcnt  of  the 
heterosis  concept,  we  must  first  consider  why  these  methods  were  used  by 
various  workers  and  how  their  use  affected  the  experimental  conclusions. 

Dr.  William  J.  Beal,  of  Michigan  Agricultural  College,  apparently  was 


16  GEORGE  HARRISON  SHULL 

the  first  to  make  extensive  use  of  controlled  cross-pollination  in  the  breed- 
ing of  corn.  Beal  was  a  student  of  Asa  Gray  from  1862  to  1865,  when  the 
latter  was  in  active  correspondence  with  Charles  Darwin.  Darwin  was  be- 
ginning the  studies  on  cross-  and  self-fertilization,  which  were  reported  in 
1877  in  an  important  book  on  the  subject.  It  has  been  thought  that  Darwin's 
views  on  the  significance  of  crossbreeding  may  have  been  instrumental  in 
inciting  and  guiding  Beal's  experiments  in  the  crossing  of  corn.  There  seems 
to  be  no  supporting  evidence,  however,  for  such  a  surmise. 

Beal's  lectures  before  various  farmers'  institutes  stressed  the  importance 
of  being  able  to  control  the  source  of  the  pollen,  so  that  the  choice  of  good 
ears  in  the  breeding  program  would  not  be  nullified  by  pollen  from  barren 
stalks  and  other  plants  of  inferior  yielding  capacity.  On  this  point  Professor 
Perry  Greeley  Holden,  for  several  years  assistant  to  Dr.  Beal,  has  stated  that 
controlled  parentage,  not  heterosis,  was  the  aim  of  the  corn  breeding  pro- 
gram at  Michigan  and  at  Illinois  before  1900. 

In  1895  Holden  was  invited  by  Eugene  Davenport  to  become  professor  of 

agricultural  physics  at  the  University  of  Illinois.  Davenport  also  had  served 

for  several  years  as  assistant  to  Dr.  Beal  at  Michigan.  Like  Holden,  he  was 

very  enthusiastic  about  the  importance  of  Beal's  program,  so  it  was  natural 

that  Davenport  and  Holden  should  agree  that  corn  improvement  be  a  major 

undertaking  of  Holden's  new  department  at  the  University  of  Illinois.  On 

initiating  this  work  at  the  University  of  Illinois,  they  learned  that  Morrow 

and  Gardner  already  had  tested  Beal's  variety  crossing  at  Illinois  before  they 

got  there,  and  with  confirmatory  results.  Concerning  the  motivation  of  all 

this  early  work,  both  at  Michigan  and  at  Illinois,  Holden  says: 

1.  Hybrid  corn  [as  we  know  it  today]  was  unknown,  not  even  dreamed  of,  previous  to 
1900.  2.  Controlled  parentage  was  the  dominant  purpose  or  object  of  this  early  corn  improve- 
ment work. 

Holden  thus  makes  it  clear  that  while  heterosis  was  at  play  in  all  of  this  early 
work,  it  was  not  the  result  of,  nor  did  it  result  in,  a  heterosis  concept. 

I  refer  next  to  the  matter  of  inbreeding,  which  some  writers  have  confused 
with  the  crossing  that  has  brought  the  benefits  of  heterosis.  Enough  selfing 
had  been  done  with  corn  prior  to  1900  to  convince  all  of  those  who  had  had 
experience  with  it  that  it  resulted  in  notable  deterioration.  The  results  of  these 
early  observations  are  aptly  summed  up  by  Holden  in  the  statement  that 
"Inbreeding  proved  to  be  disastrous — the  enemy  of  vigor  and  yield."  No- 
where, so  far  as  I  have  been  able  to  determine,  did  any  of  the  early  inbreed- 
ers  discover  or  conceive  of  the  establishment  of  permanently  viable  pure  lines 
as  even  a  secondary  effect  of  inbreeding. 

In  1898  A.  D.  Shamel,  then  a  Junior  in  the  University  of  Illinois,  offered 
himself  to  Holden  as  a  volunteer  assistant  without  pay.  He  did  so  well  that 
when  Holden  severed  his  connection  with  the  University  in  1900,  Shamel 
was  appointed  his  successor,  and  continued  in  this  capacity  until  1902.  He 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  17 

then  transferred  to  the  United  States  Department  of  Agriculture  and  did  no 
further  work  with  corn.  In  Shamel's  final  report  of  his  own  corn  experiments 
(1905),  he  laid  no  stress  on  the  positive  gains  which  resulted  from  cross- 
breeding, but  only  on  the  injurious  effects  of  inbreeding.  His  "frame  of  ref- 
erence" was  the  normally  vigorous  crossbred  (open-pollinated)  corn,  and  the 
relation  between  self-fertilized  and  cross-fertilized  corn  was  that  of  something 
subtracted  from  the  crossbred  level,  not  something  added  to  the  inbred  level. 
The  prime  objective  in  a  breeding  program,  he  said,  "is  the  prevention  of  the 
injurious  effects  of  cross-fertilization  between  nearly  related  plants  or  in- 
breeding." In  summing  up  the  whole  matter  he  said: 

In  general,  ...  it  would  seem  that  the  improvement  of  our  crops  can  be  most  rapidly 
effected  with  permanent  beneficial  results  by  following  the  practice  of  inbreeding,  or  cross- 
ing, to  the  degree  in  which  these  methods  of  fertilization  are  found  to  exist  naturally  in  the 
kind  of  plant  under  consideration. 

This  means,  for  corn,  practically  no  self-fertilization  at  all,  and  makes  it 
obvious  that,  at  least  for  Shamel,  the  heterosis  concept  had  not  yet  arrived. 

Edward  Murray  East  was  associated  with  the  corn  work  at  the  University 
of  Illinois,  off  and  on,  from  1900  to  1905.  He  worked  mainly  in  the  role  of  ana- 
lytical chemist  in  connection  with  the  breeding  program  of  C  G.  Hopkins 
and  L.  H.  Smith.  He  must  have  been  familiar  with  the  inbreeding  work  of 
Shamel,  if  not  with  that  of  Holden.  It  is  generally  understood  that  he  did 
no  self-fertilizing  of  corn  himself,  until  after  he  transferred  to  the  Connecti- 
cut Agricultural  Experiment  Station  in  1905.  Some  of  his  inbred  lines  at 
Connecticut  may  have  had  the  inbreeding  work  at  Illinois  back  of  them,  as 
he  secured  samples  of  seeds  of  the  Illinois  inbreds  sent  to  him  by  Dr.  H.  H. 
Love,  who  assisted  him  for  one  year  and  succeeded  him  at  Illinois.  But  ac- 
cording to  his  subsequently  published  records  these  older  inbred  lines  did  not 
enter  to  any  important  extent  into  his  studies  in  Connecticut. 

As  reported  in  Inbreeding  and  Outbreeding  (East  and  Jones,  pp.  123,  124), 
"The  original  experiment  began  with  four  individual  plants  obtained  from 
seed  of  a  commercial  variety  grown  in  Illinois  known  as  Leaming  Dent." 
Table  III  (p.  124)  presents  the  data  for  these  four  lines  for  the  successive 
years  from  1905  to  1917,  and  clearly  indicates  that  the  selfing  was  first  made 
in  1905.  East's  work  is  so  adequately  presented  in  this  excellent  book  that  it 
seems  unnecessary  to  comment  on  it  further  here  except  to  recall  that,  as 
shown  by  his  own  specific  statements,  my  paper  on  "The  composition  of  a 
field  of  maize"  gave  him  the  viewpoint  that  made  just  the  difference  between 
repeated  observations  of  heterosis  and  the  heterosis  concept.  In  proof  of  this 
we  have  not  only  his  letter  to  me,  dated  February  12, 1908,  in  which  he  says: 
"Since  studying  your  paper,  I  agree  entirely  with  your  conclusion,  and  won- 
der why  I  have  been  so  stupid  as  not  to  see  the  fact  myself";  but  we  also 
have  the  published  statements  of  his  views  just  before  and  just  after  the 
publication  of  my  paper.  Thus,  we  read  in  his  Conn.  Agr.  Exp.  Sta.  Bull.  158, 


18  GEORGE  HARRISON  SHULL 

"The  relation  of  certain  biological  principles  to  plant  breeding,"  which  was 
published  in  1907,  only  a  few  months  before  I  read  my  paper  in  his  presence 
in  Washington,  D.C.,  what  seems  like  an  echo  of  the  final  conclusion  of 
Shamel,  above  cited.  In  this  bulletin  East  urged  that  "corn  breeders  should 
discard  the  idea  of  forcing  improvement  along  paths  where  nothing  has  been 
provided  by  nature,"  specifically  rejecting  a  program  of  isolation  of  uniform 
types  because  of  a  "fear  of  the  dangers  of  inbreeding,"  adding  that  he  was 
"not  able  to  give  a  reason  for  this  belief  beyond  the  common  credence  of  the 
detrimental  effects  of  inbreeding."  He  returned  to  this  problem  of  the  de- 
terioration due  to  inbreeding  in  his  Annual  Report  to  the  Conn.  Agr.  Exp. 
Sta.  for  1907-8,  prepared  in  19C8,  with  my  paper  before  him.  In  this  report 
he  says: 

I  thought  that  this  deterioration  was  generally  due  to  the  establishment  and  enhance- 
ment of  poor  qualities  common  to  the  strain.  ...  A  recent  paper  by  Dr.  George  H.  Shull 
("The  composition  of  a  field  of  maize")  has  given,  I  believe,  the  correct  interpretation  of 
this  vexed  question.  His  idea,  although  clearly  and  reasonably  developed,  was  supported 
by  few  data;  but  as  my  own  experience  and  experiments  of  many  others  are  most  logically 
interpreted  in  accordance  with  his  conclusions,  I  wish  here  to  discuss  some  corroboratory 
evidence. 

We  have  thus  far  failed  to  recognize  the  existence  of  a  general  heterosis 
concept  among  plant  breeders,  prior  to  the  reading  of  my  paper  on  "The 
composition  of  a  field  of  maize"  in  January,  1908,  even  when  they  were  using 
the  methods  of  inbreeding  and  controlled  crossing  in  which  such  a  concept 
could  have  developed.  1  must  mention,  however,  a  near  approach  to  such  a 
concept  from  the  side  of  the  animal  breeders.  Before  the  American  Breeders' 
Association,  meeting  in  Columbus,  Ohio,  1907,  Quintus  I.  Simpson,  an  ani- 
mal breeder  from  Bear  Creek  Farm,  Palmer,  Illinois,  read  a  paper  which 
definitely  recognized  hybridization  as  a  potent  source  of  major  economic 
gains  beyond  what  could  be  secured  from  the  pure  breeds.  The  title  of  his 
paper,  "Rejuvenation  by  hybridization,"  is  more  suggestive  of  the  views  of 
Thomas  Andrew  Knight  than  of  the  current  students  of  heterosis,  but  the 
distinction  seems  to  me  to  be  very  tenuous  indeed. 

Although  I  listened  with  great  interest  to  Simpson's  paper,  I  do  not  think 
that  I  recognized  any  direct  applications  of  his  views  to  my  results  with 
maize.  I  was  working  within  the  material  of  a  single  strain  of  a  single  species, 
and  not  with  the  hybridizations  between  different  well  established  breeds  to 
the  superiority  of  whose  hybrids  Simpson  called  attention. 

Students  may  make  varying  estimates  as  to  how  closely  the  work  of  men 
to  whom  I  have  referred  approached  the  heterosis  concept  as  we  understand 
it  today.  But  there  can  be  no  doubt  that  there  was  a  beginning  of  this  concept 
in  the  course  of  my  own  experiments  with  corn.  At  the  beginning  of  1907  I 
had  not  the  slightest  inkling  of  such  a  concept.  By  the  end  of  1907  I  had 
written  the  paper  that  brought  such  concept  clearly  into  recognition.  At  that 
time  I  knew  nothing  of  the  work  of  Beal,  Holden,  Morrow  and  Gardner, 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  19 

McCluer,  Shamel  or  East,  in  the  selling  and  crossing  of  the  maize  plant. 
This  will  become  obvious  as  I  explain  the  motivation  and  plan  of  procedure 
of  my  corn  experiments. 

Upon  arriving  at  the  Station  for  Experimental  Evolution  at  Cold  Spring 
Harbor  on  May  2,  1904, 1  found  the  laboratory  building  unfinished.  It  was  in 
fact  not  ready  for  occupation  until  the  following  November.  The  potentially 
arable  portion  of  the  grounds  was  in  part  a  swampy  area  in  need  of  effective 
provision  for  drainage.  The  rest  had  been  at  one  time  used  as  a  garden.  But 
it  had  lain  fallow  for  an  unknown  number  of  years,  and  was  covered  with  a 
heavy  sod  that  would  need  a  considerable  period  of  disintegration  before  it 
could  be  used  satisfactorily  as  an  experimental  garden.  The  total  area  avail- 
able was  about  an  acre. 

In  the  middle  of  this  small  garden  plot  was  a  group  of  lusty  young  spruce 
trees.  These  had  to  be  removed  in  order  to  use  the  area  for  experimental 
planting  the  following  spring.  The  ground  was  plowed,  disked,  and  planted 
as  soon  as  possible  to  potatoes,  corn,  sorghum,  buckwheat,  sugar  beets,  tur- 
nip beets,  and  many  kinds  of  ordinary  garden  vegetables.  None  of  them 
were  designed  as  the  beginning  of  a  genetical  experiment,  but  only  as  an  ex- 
cuse for  keeping  the  ground  properly  tilled  so  it  would  be  in  best  possible 
condition  for  use  as  an  experimental  garden  later.  Due  to  this  fact,  no  ade- 
quate record  was  made  of  the  origin  of  the  several  lots  of  seeds  which  were 
planted.  This  is  unfortunate  in  the  several  cases  in  which  some  of  these  cul- 
tures did  provide  material  for  later  experimental  use. 

There  were  two  cultures  of  corn,  one  a  white  dent,  the  other  a  Corry 
sweet  corn.  These  two  varieties  were  planted  at  the  special  request  of  Dr. 
Davenport,  who  wished  to  have  available  for  display  to  visitors  the  striking 
illustrations  of  Mendelian  segregation  of  starchy  and  sugary  grains  on  the 
single  ears  of  the  crossbred  plants.  I  planted  the  white  dent  corn  with  my 
own  hands  on  May  14, 1904,  and  must  have  known  at  the  time  that  the  grains 
came  from  a  single  ear.  Although  I  have  found  no  contemporary  record  to 
that  effect,  I  am  now  convinced  from  a  well-remembered  conversation  with 
Mrs.  Davenport,  that  this  ear  of  white  dent  corn  came  from  the  farm  of  her 
father,  Mr.  Crotty,  who  lived  near  Topeka,  Kansas. 

When  I  was  last  in  Ames,  after  almost  forty  years  of  devotion  to  other 
lines  of  genetical  experimentation,  my  memory  played  me  false  when  Profes- 
sor J.  C.  Cunningham  asked  me  about  the  source  of  the  foundation  stock  for 
my  experimental  work  with  corn,  and  I  told  him  that  my  studies  on  corn 
began  with  some  corn  I  had  purchased  in  the  local  market  as  horse  feed.  I  re- 
peated the  same  unfortunate  misstatement  to  several  other  highly  reputable 
historians  of  science.  I  deeply  regret  this  error  because  these  men  were  tr}-ing 
so  hard  to  get  the  record  straight.  My  recollection  was  restored  by  iinding 
the  statement  at  the  very  beginning  of  the  record  of  my  formal  corn  studies 


20  GEORGE  HARRISON  SHULL 

under  date  Nov.  7,  1904:  "Counted  the  rows  on  the  ears  of  White  dent  corn 
raised  in  Carnegie  garden  this  year."  In  fact,  as  I  think  of  it  now,  I  doubt 
that  I  could  have  bought  white  dent  corn  in  the  feed  market  of  Long  Island 
at  that  time. 

I  planted  the  Corry  sweet  corn  on  May  17.  On  July  18  I  bagged  the  corn 
preparatory  to  making  crosses  between  the  two  varieties.  This  crossing  was 
carried  out  on  the  Corry  sweet  on  July  25,  and  the  crosses  for  the  reciprocal 
combination  were  made  on  July  27  and  28.  These  were  the  first  controlled 
pollinations  I  ever  made  in  corn,  and  they  were  not  part  of  a  scientific  ex- 
periment. 

My  interest  in  investigating  the  effects  of  cross-  and  self-fertilization  in 
maize  arose  incidentally  in  connection  with  a  projected  experiment  with 
evening  primroses  (Oenothera)  to  determine  the  effect,  if  any,  of  these  two 
types  of  breeding  on  the  kinds  and  the  frequencies  of  occurrence  of  mutations. 
A  critic  of  De  Vries's  mutation  theory  had  urged  that  the  mutations  dis- 
covered by  De  Vries  in  Oenothera  lamarckiana  were  artifacts  produced  by 
selfing  a  species  which,  in  a  natural  state,  had  been  always  cross-fertilized.  I 
developed  a  program  to  put  this  question  to  a  crucial  test .  Then,  it  occurred  to 
me  that  it  would  be  interesting  to  run  a  parallel  experiment  to  test  the  effects 
of  crossing  and  selfing  on  the  expressions  of  a  purely  fluctuating  character. 
Since  I  had  available  this  culture  of  white  dent  maize,  I  chose  the  grain-row 
numbers  on  the  ears  of  corn  as  appropriate  material  for  such  a  study.  The 
Oenothera  problems  thus  begun,  continued  to  be  a  major  interest  throughout 
my  genetical  career,  but  it  is  not  expedient  to  pursue  them  further  here.  It  is 
important,  however,  to  keep  them  in  mind  as  a  key  to  my  motivation  in 
launching  my  studies  with  maize. 

In  this  double-barreled  exploration  of  the  genetical  effects  of  cross-fertili- 
zation versus  self-fertilization,  I  had  no  preconception  as  to  what  the  out- 
come of  these  studies  would  be  in  either  the  mutational  or  the  fluctuational 
field.  Certainly  they  involved  no  plan  for  the  demonstration  of  distinctive 
new  biotypes,  nor  any  thought  of  the  possible  economic  advantages  of  either 
method  of  breeding.  I  was  a  faithful  advocate  of  the  early  biometricians'  slo- 
gan: Ignoramus,  in  hoc  signo  lahoremus.  Until  the  middle  of  summer  of  1907, 
certainly,  I  had  no  premonition  of  the  possible  existence  of  a  heterosis  prin- 
ciple which  would  have  important  significance  either  scientifically  or  eco- 
nomically. I  was  forced  to  recognize  this  principle  by  direct  observations  of 
manifestations  in  my  cultures  which  had  not  been  anticipated,  and  there- 
fore could  not  have  been  planned  for. 

Let  us  proceed  then  to  a  description  of  my  experiments  with  corn  which 
forced  the  recognition  of  this  important  phenomenon.  The  culture  of  white 
dent  corn  which  we  had  growing,  almost  incidentally,  on  the  Station  grounds 
that  first  year,  showed  no  variations  that  seemed  to  indicate  the  presence  of 
any  segregating  characteristics.  It  appeared  to  be  ideal  material  for  the  study 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT 


21 


of  fluctuations  of  so  definite  and  easily  observed  a  quantitative  character  as 
the  number  of  the  rows  of  grains  on  the  ears.  The  crop  was  carefully  har- 
vested and  placed  in  a  crib.  On  November  7,  1904,  I  counted  the  rows  of 
grains  on  every  ear,  with  the  result  shown  in  figure  2.1.  The  524  ears  ranged 
over  the  seven  classes  from  10-rowed  to  22-rowed.  The  most  populous  classes 


10   ... 

..       3 

12   ... 

. .     93 

14   ... 

. .   201 

16   ... 

. .   153 

18   ... 

. .     58 

20   ... 

.  .     12 

22   ... 

..       4 

Total 

. .  524 

Mean, 

14.827 

±  .061 

c>, 

2.082 

±  .043 

c.  v., 

14.02 

±:  .29 

8 


10 


12  14 

NUMBER 


16  18  20 

OF   GRAIN   ROWS 


Fig.  2.1 — Frequency  curve  of  grain-rows  of  524  ears  of  white  dent  corn.  The  total  progeny 

of  presumably  a  single  ear  of  corn  received  from  the  Crotty  farm  near  Topeka,  Kansas,  and 

grown  at  the  Station  for  Experimental  Evokition  in  1904. 

were  the  14-rowed  with  a  frequency  of  201,  and  16-rowed  with  153  individual 
ears.  The  mean  was  14.85  ±  .06. 

No  photograph  nor  verbal  description  was  made  of  the  parent  ear,  since 
there  was  no  intention  at  the  time  of  its  planting  to  use  it  in  a  breeding  ex- 
periment. But  its  characteristics  must  have  been  accurately  duplicated  in  all 
of  the  crossbred  families  subsequently  grown,  as  well  as  in  most  of  the  Fi  hy- 


22 


GEORGE  HARRISON  SHULL 


brids  between  the  several  selfed  lines.  From  each  of  the  grain-row  classes, 
several  good  ears  were  saved  for  planting  in  the  spring  of  1905,  and  the  rest 
was  used  as  horse  feed. 

The  plantings  from  this  material  were  made  on  May  25,  26,  27, 1905,  again 
with  my  own  hands,  in  the  form  of  an  ear-row  planting.  Two  ears  from  each 
grain-row  class  of  the  1904  crop  were  used.  The  seeds  were  taken  from  the 
mid-region  of  each  seed  ear.  An  additional  row  was  planted  from  grains  of 
each  of  the  two  parent  ears  with  16  grain-rows.  Only  modified  basal  grains 
and  modified  distal  grains  for  the  two  halves  of  the  same  row  in  the  field 
were  used.  In  Table  2.1  these  cultures  from  modified  grains  are  indicated  by 

TABLE  2.1 

GRAIN-ROW  COUNTS  OF  PROGENIES  GROWN  IN  1905  FROM  PARENT 

EARS  SELECTED  FOR  DIFFERENT  NUMBERS  OF 

GRAIN-ROWS  IN  NOVEMBER,  1904 


Culture 

Parental 
Grain- 
Rows 

Frequencies  of  Progeny  Grain-Row  Numbers 

Numbers 

10 

12 

14 

16 

18 

20 

22 

24 

26 

Totals 

Al 

10.4* 
10  B* 
UA 
UB 
14.4 
UB 
16  A 
\6B 
16.4bt 
16.44 
1656 
16  Bp 
18.4 
18  B 
20  A 
20  B 
22  A 
22  B 
22  CJ 

8 
11 
12 

3 

7 

1 

3 

4 

3 

2 

3 

3 

'r 

55 
50 
36 
30 
32 
11 
62 
31 
7 

19 

3 

5 

12 

20 

3 

10 

2 

3 

2 

47 
57 
45 
43 
58 
47 
81 
79 
19 
18 
5 

18 
36 
33 
28 
14 
9 
9 
12 

16 
15 
10 
28 
13 
26 
44 
66 

7 
16 

8 
12 
39 
29 
38 
28 
21 
20 
32 

3 

1 

1 

4 

5 

13 

10 

14 

2 

4 

4 

11 

17 

7 

14 
14 
27 
28 
24 

129 

A2 

1 

135 

A3 

104 

\4 

108 

AS 

115 

A6 

A7and8 

A9and  10.  .. 
All, 

1 

100 

200 

1 

195 

38 

All., 

59 

A12i 

A 12.. 

23 

49 

A13 

3 

1 

108 

A14 

89 

A15 

2 

10 
19 
18 
16 

1 
2 
7 

"3" 

86 

A16 

79 

A17 

85 

A18          

2 

1 

1 
1 

81 

A19 

91 

Totals... 

61 

393 

658 

468 

203 

71 

15 

3 

2 

1,874 

*  The  significance  of  the  A  and  B  in  this  column  involved  the  plan  to  use  the  A  rows  for  selfing  and  the  B 
rows  to  be  crossed  with  mixed  pollen  of  plants  in  the  corresponding  A  rows. 

t  The  subscript  b  signifies  the  use  for  planting  of  only  the  modified  basal  grains  of  the  given  ear;  and  the  sub- 
script p  refers  to  the  planting  only  of  modified  grains  at  the  "point"  or  distal  end  of  the  ear. 

t  C  represents  an  added  row  grown  to  increase  the  probability  of  finding  ears  with  still  higher  numbers  of 
grain-rows. 

Ab  and  Bb  for  the  basal  grains,  and  Ap  and  Bp  for  the  modified  "point" 
grains.  A  second  row  was  planted  from  each  of  the  two  chosen  ears  having  16 
grain-rows,  and  these  additional  rows  (A8  and  AlO)  were  detasseled,  begin- 
ning July  24,  1905,  and  received  pollen  from  the  intact  plants  in  the  corre- 
sponding rows  (A7  and  A9)  beside  them. 

In  harvesting  these  two  pairs  of  rows,  one  detasseled,  the  other  intact,  the 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  23 

two  rows  from  the  same  parent  ear,  through  an  oversight,  were  not  kept 
separate.  No  further  detasseling  was  done.  Since  the  self-fertilized  plants 
could  not  be  detasseled  and  still  utilized  for  selling,  the  method  of  controlling 
cross-fertilization  by  detasseling  would  prove  a  distorting  factor  in  comparing 
the  effects  of  selling  and  crossing. 

Consequently,  no  detasseling  was  practiced  in  any  of  my  subsequent  ex- 
perimental work  with  corn,  but  every  pollination  was  controlled  by  bagging 
with  glassine  bags  and  manipulation  by  hand.  The  bags  were  tied  in  place 
by  ordinary  white  wrapping-cord  passed  once  around  and  tied  with  a  loop 
for  easy  detachment.  Each  plant  was  labeled  at  the  time  of  crossing  with  a 
wired  tree-label  attached  to  the  stalk  at  the  height  of  the  operator's  eyes, 
and  marked  with  the  exact  identification  of  the  plant  to  which  it  was  attached 
and  the  source  of  the  pollen  which  had  been  applied.  On  harvesting  these 
hand-pollinated  ears,  the  label  was  removed  from  the  plant  and  attached 
securely  to  the  ear,  thus  assuring  that  the  ear  and  its  label  would  remain 
permanently  associated.  A  third  row  (A  19)  from  an  ear  having  22  grain-rows 
was  added  to  improve  the  chances  of  finding  ears  with  still  higher  numbers 
of  grain-rows. 

In  November,  1905,  these  19  pedigree  cultures  were  carefully  harvested 
by  my  own  hands  and  the  grain-rows  counted,  with  the  results  tabulated  in 
Table  2.1. 

The  only  observation  noted  on  these  1905  cultures  was  that  there  was  no 
clear  indication  of  mutations  or  segregations  of  any  kind,  but  the  aspect  of 
the  field  was  that  of  any  ordinarily  uniform  field  of  corn.  Row  counts  did 
show  the  expected  indication  of  Galtonian  regression,  in  that  the  parents 
with  low  numbers  of  grain-rows  produced  progenies  having  lower  numbers  of 
grain-rows  than  did  the  ears  having  higher  than  average  numbers  of  grain- 
rows.  Thus,  the  two  ears  with  10  rows  of  grains  each  had  the  average  of  13.2 
rows  of  grains  on  their  progeny  ears.  The  two  20-rowed  ears  showed  an  aver- 
age of  15.5  rows  of  grains  on  their  progeny  ears.  The  three  22-rowed  parent 
ears  produced  progenies  with  an  average  of  17.5  rows  of  grains. 

The  same  general  plan  was  followed  in  1906,  except  that  the  pollen  for 
the  crossbred  cultures  was  no  longer  taken  from  the  plants  set  aside  for 
selfing.  The  reason  for  this  change,  as  specifically  stated  in  my  notes  written 
at  the  end  of  the  1906  season,  being  "to  avoid  the  deleterious  effects  of  self- 
fertilization  in  the  cross-fertilized  series."  This  indicated  that  at  the  end  of 
19C6  I  had  only  the  concept  held  by  Holden,  Shamel,  East,  and  all  other 
corn  breeders  who  had  had  experience  with  the  selfing  of  maize — that  selfing 
has  deleterious  effects,  not  that  crossing  has  advantageous  effects  other  than 
the  simple  avoidance  of  the  deleterious  effects  of  selfing. 

The  new  method  of  handling  the  crossbred  cultures  was  to  divide  each 
such  culture  by  a  marker  set  at  the  midpoint  of  the  row.  All  the  plants  in 
these  rows  were  bagged.  Mixed  {)ollen  from  the  plants  in  the  first  half  of  the 


24  GEORGE  HARRISON  SHULL 

row  was  collected  and  applied  at  the  appropriate  time  to  the  silks  of  all  the 
plants  in  the  second  half  of  the  row.  Then  the  mixed  pollen  from  the  plants 
in  the  second  half  of  the  row  was  applied  in  turn  to  the  silks  of  all  the  plants 
in  the  first  half  of  the  row.  It  was  realized  that  this  still  involved  a  con- 
siderable degree  of  inbreeding,  but  it  seemed  about  the  only  way  of  carrymg 
on  a  continuing  program  of  crossing  while  still  keeping  the  breedmg  com- 
pletely under  the  operator's  control. 

Two  major  observations  made  on  the  1906  crop  were:  (1)  that  every  one 
of  the  seven  families  from  selfed  parents  could  be  readily  detected  by  their 
less  height,  more  slender  stalks,  and  greater  susceptibility  to  theattack  ot 
Ustila2o  maydis.  When  the  ears  were  harvested  each  lot  was  weighed  and 
it  was  found  that  cross-fertilized  rows  produced  on  an  average  about  three 
times  as  much  grain  as  the  self-fertilized.  (2)  The  family  A3,  from  a  self- 
fertilized  ear  having  12  grain-rows,  was  practically  all  flint  corn,  showing  that 
to  be  probably  recessive.  This  occurrence  of  a  rather  obvious  segregation  in 
the  1906  crop  remained  at  the  end  of  the  season  only  an  isolated  observation 
which  led  to  no  generalization.  From  the  fall  of  1905  until  his  retirement 
in  1943  Charles  Leo  Macy  assisted  me  in  many  of  the  technical  details  of  my 
experimental  cultures.  While  I  handled  the  planning  and  breeding  operations 
as  well  as  the  actual  poUinations,  Macy  prepared  the  plants  for  selfing  and 
crossing,  and  counted  the  grain-rows  and  weighed  the  ear  corn.  The  results 
of  these  counts  for  the  1906  crop  are  given  in  Table  2.2.  _ 

The  following  quotation  from  my  notebook  seems  justified  here,  since  it 
includes  the  first  formulation  of  the  considerations  and  conclusions  which 
appeared  in  my  report  to  the  American  Breeders'  Association  in  1908,  on 
"The  composition  of  a  field  of  maize": 

^-       A    c;,,  1007  p';  in  1906)   namely  each  self -fertilized  row  was 

The  same  plan  was  continued,  ^i"  l^^^;^^  ^^^^^^^  row  was  divided  in  half, 

the  offspring  of  a  smgle  self-fertilized  far,  and  each  cross  termzea  ^^^^ 

each  half  coming  from  a  single  cross-fertilized  e^^^' f^^^.^f"  ^"J^'e  ^Vhlr  ear  coming  from  the 
the  first  half  of  the  corresponding  row  of  the  precedmg  >  ear,  the  other  ear  c  n     g 

second  half.  ...  .     mnA   tv,o  coif  fprtili/pd  rows  being  invariably 

The  obvious  results  were  the  same  as  in  ^^O^,  the  self -fert^ed    o^^  s         g  ..^ 

smaller  and  weaker  than  the  corresponding  """f  "^^  ^iliz'-^j^^f  l^^^  upon  me 

evidence  on  the  self-fertilized   A  v^ry  df  erent  explana  ion  of^  h^  P^.^.^^ 

by  the  fact  that  the  several  self-  ertlllzedro^^s  differ  rom  each  ^^^^^^^^  elementary 

morphological  characteristics,  thusindicatng  that  the,    be^^^^  ^^^^^  ^^.^.^^  ^^^, 

strains.  The  same  point  appeared  last  >ear  in  the  c^se  oi  tne  ^      ^^^^ 

almost  a  uniform  flint  corn,  but  the  ^^^'^^^""^^f.^^^J^Z^^SSr^^^  that  my  corn- 
It  now  appears  that  self-fertilization  simply  ^^'^f.  ^  P^"^ >  j^^'  but  between  pure  strains 
parisons  are  not  properly  between  "^^J^^f  .^/^"^Jj^'^j^^^^^ 

and  their  hybrids;  and  that  a  well  '•^g^^^^^f  ,^^^  ^  °4e^^  ^  of 'corn  must  have  as 

its  ?b.-t?^rn!?inraTc^  ^^s^ h^d  ^^^^^^^^^^^^^  to\e  most  vigorous  and 

productive  and  give  all  desirable  qualities  of  ear  and  grain. 

The  ideas  in  this  quotation  represent  a  discovery  in  complete  disagree- 
ment with  my  preconception  that  my  white  dent  foundation  stock,  which 
had  been  the  progeny  of  a  single  ear,  was  essentially  a  ^^^^'^^^^y  f'^l?^'^^' 
I  had  before  me  seven  distinct  biotypes,  clearly  distinguishable  m  their  sev- 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT 


25 


eral  morphological  characteristics.  They  had  been  derived  from  seven  sepa- 
rate self-pollinations  of  sibs  in  a  family  which  I  had  reason  to  think  was 
genetically  homogeneous.  This  could  not  fail  to  make  a  great  imj)ression. 
Had  these  several  pure-bred  self-fertilized  strains  come  from  different 
breeders  and  from  more  or  less  disconnected  experiments,  as  did  the  selfed 

TABLE  2.2 

GRAIN-ROW  COUNTS  AND  YIELDS  OF  EAR  CORN  IN  CULTURES  OF 

WHITE  DENT  MAIZE  GROWN  AT  THE  STATION  FOR 

EXPERIMENTAL  EVOLUTION  IN  1906 


Culture 
Numbers 

Parental 
Grain-Rows 

Frequencies  of  Progeny 
Grain-Row  Numbers 

To- 
tals 

Weights 
Lbs.Av. 

Yield 
Bu./A. 

8 

10 

12 

14 

16 

18 

20 

22 

24 

26 

28 

30 

32 

Al.l 

10  selfed 
10  crossed 
10  crossed 
12  selfed 
12  crossed 
12  crossed 
14  selfed 
14  crossed 
14  crossed 
16  selfed 
16  crossed 
16  crossed 
16(22)X10 
16^  crossed 
16j,  crossed 

16p  crossed 

16 J,  crossed 

18  selfed 
18  open-pol. 
20  selfed 
20  crossed 
20  crossed 
22  selfed 
24  crossed 
24  open-pol. 
26  open-pol. 
26  open-pol. 
18  crossed 
18(22)X10 
14(22)  XIO 

4 
'5' 

36 
3 
2 

13 

1 
"1 

62 

32 

26 

40 

13 

16 

12 

6 

6 

8 

14 

5 

8 

9 

20 

7 

10 

2 

14 
25 
29 
19 
26 
34 
41 
28 
17 
17 
16 
16 
23 
28 
23 

39 

22 

8 
16 
11 

2 
3 
4 
4 
12 
1 

1 

7 
11 

4 
12 

9 
34 
18 
19 
28 
15 
28 
22 
20 
15 

18 

18 

6 
29 
23 

8 
17 
10 
11 
14 

8 

5 

17 
20 
22 

117 
68 
69 
82 

581 
61/ 
107 
59\ 
58J 
74 
471 
61/ 
71 
60 
63 

74 

57 

26 
94 

72 

,t?l 

55 
91 
66 
60 
68 
78 
72 
92 

A2.2i 

1 

1 

1 

6 

1 
15 

7 
12 
17 

1 

11 
11 

3 

9 

5 

5 

18 
18 
21 
20 
17 
25 
17 
11 

9 
25 

6 
26 

A2.22 

A3.3 

21.6 
65.8 
33.6 
61.3 
29.6 
58.3 
22.1 

37.7 

A4.4i 

1 

A4.42 

78.9 

AS.5 

4 

44  9 

A6.6i 

A6.62 

4 
4 

1 
1 
5 

74.5 

A7.7 

59.1 

A9.8i 

.'^9.82 

77.1 

A19.9 

44.5 

A12i.lOi 

AI22.IO2 

A12i/"' 

A122i"' 

A13.12 

1 

2 

3 
19 
10 

13 
13 
13 
24 
11 
17 
14 
21 

2 
9 
4 
5 
4 
7 

18 
6 
6 

19 
7 

9.6 
58  3 
23  6 

56.3 

24.1 
57.3 
32.6 
34.6 
40  6 

52  9 

A14.13 

1 

1 

88.5 

A15.14 

6 

46  9 

A16.15i 

1 

A16.152 

75. 1 

A17.16 

1 

3 
3 
2 

10 

13 

2 

62  4 

A18.17 

4 

89.9 

A19.18 

1 

3 

70.6 

A19.19 

7 

5 

82  4 

A18.20 

85.4 

A16  21 

"i 

1 
16 
11 

5 
29 
31 

A19. 22    .  . 

A19. 23 

46   1 

71.6 

Totals 

9 

58 

334 

543 

469 

323 

183 

89 

36 

17 

3 

2 

1 

2,067 

lines  available  to  Dr.  East,  the  observation  that  they  showed  themselves  to 
be  genetically  distinguishable  biotypes  would  have  given  no  cause  for  the 
special  conclusions  I  drew  from  them.  It  would  have  been  strange,  indeed, 
if  strains  thus  derived  from  heterogeneous  sources  had  not  been  genetically 
different,  one  from  another. 

Comparison  of  the  results  for  1907,  presented  in  Table  2.3,  with  those  for 
1906  in  Table  2.2,  shows  a  heavy  accentuation  of  grain-row  classes  8  and  10 
and  a  marked  decrease  in  classes  18  to  20,  inclusive.  There  was  also  a  sig- 
nificant increase  in  all  higher  classes,  with  further  extension  of  the  range  from 
a  maximum  of  32  to  about  40.  The  increase  in  the  frequencies  of  the  low 


26 


GEORGE  HARRISON  SHULL 


grain-row  classes  was  attributed  in  part  to  the  fact  that  the  1907  season  had 
seemed  less  favorable  in  general  than  1906. 

It  was  also  noted,  as  a  possible  contributory  condition,  that  this  was  the 
third  season  in  which  this  corn  was  grown  on  the  same  area  north  of  the 
laboratory  building,  and  that  ''the  yield  may  have  been  lessened  by  the 
gradual  accumulation  of  injurious  substances  in  the  soil."  The  fact  that  the 


Fig.  2.2 — Young  corn  cultures  growing  in  East  Garden  of  the  Station  for  Experimental 
Evolution  in  1911,  illustrating  that  no  two  were  alike  despite  their  descent  from  a  single  ear 
of  1904  by  meticulously  controlled  pollinations  that  precluded  the  introduction  of  pollen 

from  any  other  strain  of  corn. 


average  grain-row  numbers  were  not  significantly  different  in  the  two  years — 
15.8  in  1906,  16.0  in  1907 — in  fact  a  trifle  higher  in  what  was  thought  to  have 
been  the  poorer  year,  does  not  seem  to  support  these  suggested  explanations 
of  the  observed  differences  of  distribution  in  the  two  years. 

My  contemporaneous  notes  proposed  an  additional  explanation,  namely, 
that  "each  successive  generation  of  close  inbreeding  still  further  reduces  the 
strains  to  their  simple  constituent  biotypes,  and  as  these  are  weaker  than 
hybrid  combinations,  this  too  would  tend  to  lessen  the  vigor,  and  this 
lessened  vigor  might  readily  be  evidenced  by  a  decrease  in  the  average  num- 
ber of  [grain-] rows  and  the  total  number  of  ears  in  the  crop." 

If  we  accept  this  latter  suggestion  as  valid,  it  is  clear  that  the  occurrence 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT 


27 


of  essentially  the  same  average  numbers  of  grain-rows  in  the  two  years  gives 
only  a  specious  indication  of  the  relative  climatic  and  soil  effectiveness  in 
these  two  seasons.  It  must  mean  simjjly  that  the  diminution  of  grain-row 
numbers  produced  by  increasing  homozygosity  happened  to  be  balanced  by 
the  increased  frequencies  in  the  higher  classes,  produced  by  the  gradual  ac- 
cumulation by  selection  of  more  potent  hybrid  combinations. 

TABLE  2.3 

GRAIN-ROW  COUNTS  AND  HEIGHTS  OF  PLANTS  IN 
THE  CULTURES  OF  1907 


Pedigree 

Grain-Rows 
of  p.\rents 

Frequencies 

DF  Progeny  Grain- Row  Numbers 

To- 
tals 

Av. 
Ht 

Numbers 

8 

10 

12 

14 

16 

18 

20 

22 

24 

26 

28 

30 

32 

34 

36 

38 

40 

IN 

Ft. 

Bl.l 

10  selfed 
10  crossed 
10  crossed 
8  selfed 
12  selfed 
12  crossed 
12  crossed 
14  selfed 
14  crossed 
14  crossed 
16  selfed 
16  crossed 
16  crossed 

16    crossed 
I61  crossed 
18  selfed 
18  open-pol. 
20  selfed 
20  crossed 
20  crossed 
22  selfed 
22  crossed 
20  crossed 
24  crossed 
32  open-pol. 
Branched  ear 
30  open-pol. 
Branched  ear 
16  crossed 
24  selfed 
26  selfed 

20 

2 

6 

23 

10 

"1 

25 
22 
28 
48 
21 
1 

'  '4 

20 
21 
18 
17 
18 
16 

7 
23 

5 

5 
19 

2 

2 

5 
5 

1 

68 

501 

57/ 

88 

54 

52 

44 

73 

48! 

43/ 

55 

371 

36/ 

311 

37/ 

15 

71 

52 

631 

52/ 

45 

63 

62 

50 

41 

58 

58 

49 

64 

36 

12 

7.25 

B2i.2i 

B22.22 

9.00 

B1.3 

7.63 

B3.4 

4 
19 
14 
29 
15 
18 
26 
8 
9 

13 
15 

1 
13 
14 
15 
19 
19 

9 
14 
15 

10 

9 

5 

22 

21 

16 

10 

6 

9 

7 

1 

3 

9 

5 

14 

14 

5 

6.25 

B4i.Si 

3 

7 

1 
7 
7 

8  00 

B42.S2 

1 
1 
1 
1 

B5.6 

8.50 

B6i   7i     .    . 

1 
1 

B62.72 

9  67 

B7.8 

8  25 

B8i.9i 

7 
9 

1 

1 

7 

15 

11 

18 

13 

9 

17 

22 

7 

6 

13 

12 

17 

15 

5 

3 
3 

1 

1 

1 

3 

6 

13 

16 

17 

17 

17 

16 

8 

8 

17 

12 

17 

7 

3 

3 

B82  92     ... 

8  75 

BIO2  lOi.  .. 

6 
11 

BlOi   IO2  . 

8.67 

B12.il 

2 

1 

1 

10 

8 

8 

11 

11 

14 

8 

1 

11 

3 

6 

7 

2 

7.00 

B13   12 

9 

21 
8 
3 
4 

8  33 

B14.13...  . 

7.25 

B15i   14i... 

2 

1 
5 
4 
4 
5 
7 
1 
9 

B152.142... 

8  83 

B16.15 

7.00 

B17.16...  . 

1 
1 

3 

1 

8  67 

B15.17 

8  83 

B19.18 

4 
4 

3 

2 

9  50 

B20.19...  . 

2 

9  50 

B?  20.    .    . 

6 

2 
3 

5 

8  33 

B17  21...  . 

1 

8  33 

B?.22 

BIS. 23 

3 
6 
2 

4 
6 

B20  24 

8.00 

B20.2S...  . 

1 

2 

1 

1 

7.83 

Totals .  . 

62 

150 

204 

236 

282 

228 

189 

108 

49 

22 

6 

3 

3 

1 

1 

1 

1,545 

A  truer  measure  of  the  relative  favorableness  of  the  two  seasons  for  growth 
and  productiveness  of  these  cultures  can  be  derived  from  a  study  of  the 
middle  classes  with  12,  14,  16,  and  18  grain-rows.  These  grain-row  classes 
making  up  80  per  cent  of  the  1906  crop  and  61.5  per  cent  of  the  1907  crop, 
must  be  relatively  free  from  most  of  the  distortion  assumed  to  be  produced 
either  by  increasing  homozygosity  or  by  the  accumulation  of  the  more  po- 
tent hybrid  combinations.  If  we  average  these  four  grain-row  classes  by  them- 
selves for  the  two  years,  we  find  that  in  1906  their  average  was  15.5  grain- 
rows,  and  for  1907  only  15.0,  thus  agreeing  with  my  general  impression 
that  1907  was  the  less  favorable  year. 

With  the  fundamental  change  in  my  understanding  of  the  nature  of  my 
corn  population  came  a  reorientation  of  the  experiment.  I  found  myself  at 


28  GEORGE  HARRISON  SHULL 

the  end  of  1907  only  ready  to  make  a  beginning  on  the  problems  of  the  rela- 
tionship between  pure  lines  and  their  hybrids,  which  I  now  saw  was  the  cru- 
cial field  that  needed  exploration. 

As  a  first  step  in  that  direction,  but  without  as  yet  a  full  comprehension 
of  its  importance,  I  made  in  July,  1907,  pollinations  between  plants  of  C4, 
which  I  later  designated  "Strain  A,"  and  a  plant  of  C6,  which  later  became 
my  "Strain  B."  I  also  made  two  sib  crosses  within  these  two  strains.  The 
cross  of  Strain  A  X  Strain  B,  which  gave  rise  in  1908  to  Fi  family,  D9,  in- 
volved an  8-rowed  ear  of  the  former  strain  (from  an  original  selection  for 
12  grain-rows)  and  a  12-rowed  ear  of  Strain  B  which  had  originated  in  a  selec- 
tion for  14  grain-rows.  The  near-reciprocal  cross  (Fi  family,  D13)  resulted 
from  the  application  of  pollen  from  a  12-rowed  plant  of  Strain  A  to  silks  of 
the  same  plant  of  Strain  B,  which  supplied  the  pollen  for  the  near- reciprocal 
cross. 

At  the  time  when  these  two  near-reciprocal  crosses  were  made  between 
Strains  A  and  B,  the  truth  had  not  yet  dawned  upon  me  that  I  should  do  the 
same  with  all  of  my  other  selfed  families.  Aside  from  these  two  sets  of  crosses, 
the  handling  of  the  cultures  was  the  same  as  in  previous  years.  The  results 
of  the  grain- row  counts  are  given  in  Table  2.4.  Unfortunately,  there  was  con- 
siderable damage  from  crows,  and  failures  to  germinate  for  unknown  reasons. 
The  missing  hills  were  replanted  on  June  8, 1908,  and  all  of  the  new  plantings 
made  on  this  date  seem  to  have  reached  maturity.  To  overcome  the  suggested 
deteriorating  effect  of  soil  depletion,  the  cultures  were  grown  this  year  on  the 
area  east  of  the  laboratory  building  (occasionally  referred  to  in  subsequent 
notes  as  "Fast  Garden"). 

In  summarizing  the  results  for  the  year  1908,  it  may  be  noted  first  that 
the  tendency  to  concentrate  the  frequencies  of  the  grain-rows  in  the  extremes 
of  the  range,  at  the  expense  of  those  in  the  middle,  has  continued  strongly. 
As  before,  the  most  noteworthy  concentration  is  at  the  lower  extreme.  All 
classes  below  16  are  considerably  stronger  in  1908  than  in  1907  and  the 
maximum  frequency  is  now  on  12  instead  of  16.  This  is  in  part  due  to  the  fact 
that  several  of  the  lower-class  families  were  grown  in  duplicate.  Between 
classes  14  and  26  the  relative  strength  of  the  classes  was  lessened  in  1908. 
Above  class  24  the  frequencies  were  increased,  there  being  84  ears  above 
class  24  in  1908  and  only  the  equivalent  of  about  50  in  the  same  region  in 
1907,  when  raised  to  the  same  total  number.  The  highest  number  of  grain- 
rows  noted  was  42. 

The  important  new  features  brought  in  by  the  near-reciprocal  crosses  be- 
tween Strain  A  and  Strain  B  and  a  sib  cross  in  Strain  A  are  presented  in  my 
report  to  the  American  Breeders'  Association  at  Columbia,  Mo.,  in  January, 
1909,  on  "A  pure  line  method  in  corn  breeding."  I  find  a  discrepancy  in  that 
the  78  ears  produced  by  the  sib  cross  weighed  only  16.25  pounds  instead  of 
16.5,  as  stated  in  my  1909  paper.  Whether  by  an  oversight  or  intentionally, 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  29 

I  cannot  now  determine,  the  corresponding  sib  crosses  in  Strain  B  were  not 
included  in  my  1909  report.  The  results  were  essentially  the  same  as  were  re- 
ported for  the  sib  cross  in  Strain  A.  Selfed  Strain  B  (see  Table  2.4,  family 
C6.ll)  showed  average  heights  of  plants  2.3  meters,  and  yielded  66  ears 
weighing  13.0  pounds.  The  two  sib  crosses  produced  plants  2.5  meters  tall 
and  yielded  89  ears  weighing  28.5  pounds.  Distribution  of  the  grain-row 
frequencies  was  closely  similar  in  selfed  and  in  sib-crossed  Strain  B,  but  sig- 
nificantly higher  in  the  latter: 

_,      .  Totals  Averages 

Grain-rows 10  12  14  16  18 

Selfed 2  20  26  17  1         66         1.3  8 

Sib-crossed 3  15  45  18  8        89         14.2 

There  was  abundant  evidence  that  the  sib  crosses  showed  a  greatly  re- 
stricted advantage  over  self-fertilization.  It  was  also  clearly  indicated  that 

TABLE  2.4 

GRAIN-ROW  COUNTS,  HEIGHTS,  AND  YIELDS  OF 
WHITE  DENT  MAIZE  GROWN  IN  1908 


Pedigree 

Gr.^in-Rows 

of  p.4rents 

Frequencies  of  Progeny 
Grain-Row  Numbers 

To- 
tals 

Av. 

Hts. 

IN 

Dm. 

Wts. 

IN 

Lbs. 

Yield 

Bu./ 

A. 

8 

10 

12 

14 

16 

17 
11 

6 
11 

3 
28 
25 
13 
34 
16 
14 
39 

4 

6 
17 

9 
22 

3 
tvs  t 

1 

14 
20 

323 

18 

"1 

1 

5 

3 

1 

2 

2 

1 

11 

4 

21 

12 

4 

23 

15 

17 

19 

20 

26 

4 

30  d 

2 

19 
31 

244 

1 
20 

"i 

6 

1 

6 
19 
10 
18 
24 
25 
21 
fBci 
5 

16 
22 

172 

22 

1 

"7 
11 
3 
13 
19 
16 

It  t( 
10 

14 

7 

91 

24 

8 

16 
21 
24 
)  coi 
10 

4 
3 

60 

26 

"2 

"4 

"7 

int; 

16 

2 
31 

28 

"i 

3 

silks 
18 

2 
24 

30 

i 

sh 
5 

6 

32 

3 

ort 
6 

9 

34 

i 

er 
9 

3d 

38 

40 

42 

C3.1 

C1.2 

C2,.3, 

C22.32 

C1.4 

C2i.5i 

C22.52 

C4.6 

C4.7 

C4.8 

C4.9 

C5i.ini.... 

C52.102.... 

C6.ll 

C6.12I..., 

C6.122 

C6.13 

C7,.14i.... 
C72.142.... 

C8.15 

C9i.l6i.... 
C92.162.... 
C13.17.... 
CI22.I81... 
C122.I82... 
C13.19.... 
C14,.20i... 

10  selfed 
8  selfed 
8  crossed 
8  crossed 
10  selfed 
10  crossed 
10  crossed 
12  selfed 
lOXsib 
8  selfed 
8X12 
12  crossed 
14  crossed 
14  selfed 
16Xsib 
12Xsib 
12X12 
14  crossed 
14  crossed 
16  selfed 
16  crossed 
16  crossed 
18  selfed 
18  crossed 
18  crossed 
20  selfed 
20  crossed 
20  crossed 
20  selfed 
22  crossed 
22  selfed 
28  crossed 
36?  selfed 
28(?)X26(?) 
Branched  ear 

open-pol. 
20  open-pol. t 

52 
51 

6 

6 
28 

9 
12 
11 

8 
65 

"2 
1 

39 

41 

29 

22 

48 

32 

18 

41 

50 

6 

19 

9 

9 

2 

2 

ii 

13 
2 

14 

12 

12 

9 

3 

32 

19 

2 

64 

31 

17 

20 

4 

11 

56 

18 

9 

31 

4 

6 

6 

6 

8 

2 

"2 

1 
2 

"5 
1 

'9 
15 
14 
26 
25 
20 
31 
28 
18 
32 
14 
18 
10 
20 
19 
15 
3 
2 

13 
3 
9 

tha 
3 

n'h 

usl 

cs. 

1 

104 
94 
51 
42 
90 
50 
33 
89 
78 
73 
92 
59 
42 
66 
47 
42 
100 
59 
43 
94 
55 
41 
77 
56 
46 
85 
58 
46 
70 
84 
92 
83 

86 

83 
93 

19.5 
19.7 

23.4 

is'o 

21.5 

'i7'0 
16.5 
16.5 
24.0 
24.5 
22.5 
23.0 
25.0 
25.0 
26.0 
25.0 
27.0 
24.4 
26.8 
25.2 
19.3 
23.5 
25.5 
21.6 
.  .  .  .* 

31.5 
22.0 
25.01 
21.0/ 

22  0 
20.81 
14.0/ 
28.0 
16.3 
12.0 
48.0 
34.8 

23  3 
13  0 
16.8 
11.8 
55 . 0 
30.0 

19  3 
31.5 
31.5 

20  0 
16.5 
31.3 
28.3 
23  0 
31.0 
24.5 
20.5 
48.3 
33.8 
43.3 

50.5 

50  0 
51.8 

43.3 
33.4 

70.7 

30.5 

59.8 

44.9 
29.8 
23.5 
74.5 
84.1 
79.1 
28.1 
50.9 
40.0 
78.6 
72.6 
64.0 
47.9 
80.8 
69.7 
30.6 
79.7 
87.7 
38.7 

CI42.2O2..  . 

76.4 

C15.21...  . 

79.2 

CI61.22.  .  . 

41.8 

C24.23.... 
CIS. 24.... 

2 

82.1 

52.4 

C25.2S.... 
C19.26.... 

Grai 

n-ro 

74.4 

C22.27.... 
C22.28.... 

1 
1 

11 
9 

83.9 

86.2 
79.5 

Totals . 

^S^ 

387 

415 

375 

10 

3 

1 

2,403 

*  The  remaining  nine  rows  were  not  measured  and  described,  "for  lack  of  time.' 
lanlil^''  '''''"'  ""'^"^  ^°"'  ^^''  "^'^^  '■*'  ''^'^^'  ^""^  ^°  '"""^  °^  ^''''"''  °^  ""'^'""^  °"'^'  "'"^  twenty-rowed  ear  was  used  for 


Fig.  2.3— Vegetative  habits  of  Strain  A  (righl)  and  Strain  B,  drawn  by  J.  Marion  Shull 
from  a  photograph  taken  in  the  summer  of  1908.  At  upper  right  typical  ears  of  these  two 
strains  {Strain  A  at  right)  and  between  them  their  reciprocal  Fi  hybrids,  each  hybrid  stand- 
ing nearest  to  its  mother  type. 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  31 

if  the  advantage  consisted  solely  of  the  effects  of  heterozygosity,  both  Strain 
A  and  Strain  B  were  still  a  good  way  from  being  homozygous,  Strain  B  being 
as  yet  more  effectively  heterozygous  than  Strain  A. 

In  the  reciprocal  crosses  between  these  nearly  homozygous  strains  A 
and  B,  we  have  our  first  opportunity  to  arrive  at  an  approximation  to  the 
actual  amount  of  heterosis.  The  most  important  new  discoveries  these 
crosses  made  possible  were:  (1)  As  a  result  of  such  a  cross  it  is  possible  to 
completely  cancel  in  a  single  year  the  accumulated  deterioration  which 
had  gradually  accrued,  although  with  lessening  annual  increments,  over  a 
period  of  several  years;  and  (2)  the  approximate  identity  of  the  results  of  the 
reciprocal  crosses  gave  assurance  that  the  amount  of  heterosis  resulting  from 
a  given  hybridization  is  a  specific  function  of  the  particular  genetical  combi- 
nation involved  in  the  cross. 

Several  new  cultures  of  yellow-  and  red-grained  corn  were  added  to  my 
experimental  field  in  1908,  but  these  will  not  be  followed  here.  They  are 
mentioned  only  because  they  were  included  in  my  numbered  pedigrees,  and 
their  omission  in  the  following  tables  leaves  a  break  in  the  series  of  numbered 
families  which  might  lead  to  some  question  as  to  the  reason  for  the  apparent 
vacancies.  The  data  from  the  1909  cultures  of  white  dent  corn  are  presented 
in  Table  2.5. 

The  families  grown  in  1909,  as  tabulated  in  Table  2.5,  fall  into  three  major 
classes:  (1)  Twelve  families  involve  continuations  of  the  original  self- fer- 
tilized lines,  whose  average  yields  range  from  18.8  to  41.2  bushels  per  acre, 
with  the  average  for  all  twelve  at  32.8  bushels  per  acre;  (2)  Twelve  are  con- 
tinuations of  crossbred  families  in  which  strictly  controlled  cross-fertiliza- 
tions were  made  with  mixtures  of  pollen  taken  from  the  other  plants  in  the 
same  crossbred  strain.  These  yielded  from  58.1  to  83.3  bushels  per  acre  with 
the  average  of  all  at  73.3  bushels  per  acre;  and  (3)  there  were  fourteen  Fi  hy- 
brid families  from  crosses  between  pairs  of  individuals  representing  two  dif- 
ferent selfed  lines.  The  yields  of  these  range  from  60.3  to  87.5  bushels  per 
acre,  the  average  for  all  fourteen  being  78.6  bushels  per  acre.  As  stated  in  my 
1910  paper,  the  three  highest  yields  of  any  of  these  cultures  were  from  the 
families  produced  by  crossing  representatives  of  different  selfed  strains  (see 
D8.13,  D8.16,  andD11.21). 

Besides  these,  there  were  two  cousin  crosses  involving  matings  between 
different  families  of  the  same  selfed  line.  These  produced,  respectively,  27.1 
and  44.6  bushels  per  acre.  One  cross  between  two  sibs  in  Strain  A  gave  26.0 
bushels  per  acre.  The  other  cross  was  two  F2  families,  each  from  crosses  with 
mixed  pollen  within  one  of  the  Fi  families  of  my  1908  cultures.  These  Fo 
families  yielded  54.2  bushels  per  acre  from  the  (A  X  B)F2,  and  70.6  from 
the  (B  X  A)F2.  These  yields  should  be  compared  with  those  of  the  corre- 
sponding Fi  families  grown  in  the  same  season,  in  which  (A  X  B)Fi  yielded 
74.9  and  83.5  bushels  in  two  different  families,  and  (B  X  A)Fi  produced 
82.6  bushels  per  acre. 


32  GEORGE  HARRISON  SHULL 

In  1910  I  was  absent  from  the  Station  for  Experimental  Evolution  during 
the  entire  summer  and  my  experiments  with  corn,  evening  primroses, 
Lychnis,  etc.,  were  continued  by  an  assistant,  R.  Catlin  Rose,  assisted  by 
Mr.  Macy,  who  carried  out  the  operations  meticulously  described  by  myself 
in  more  than  one  thousand  typewritten  lines  of  detailed  instructions. 

The  data  on  the  white  dent  corn  grown  in  1910  are  presented  here  in 

TABLE  2.5 

GRAIN-ROW  COUNTS,  HEIGHTS  OF  ST.ALKS,  AND  YIELDS  OF 
EARS  OF  WHITE  DENT  CORN  IN  1909 


Pedigree 
Numbers 

Grain-Rows 
OF  Parents 

Frequencies  of  Progeny 
Grain-Row  Numbers 

To- 
tals 

Hts. 

IN 

Dms. 

Wts. 

IN 

Lbs. 

Yield 
Bu./A. 

8 

10 

12 

14 

16 

18 

20 

22 

24 

26 

28 

30 

32 

34 

36 

38 

40 

Dt.l 

D2.2 

D3i.3i.  .  . 
D32.32.  .  .  . 

D4.4 

D4  5 

8  selfed 
8  selfed 
8  crossed 
8  crossed 
10X12 
10X14 
lOXsib 
10  selfed 
10  crossed 
10  crossed 
12  selfed 
12Xcousins 
12Xcousins 
A  selfed 
AX20 
AX22 
AXB 
AX20 
AX  16 
AXB 

(AXB)Fisibs 
12  crossed 
12  crossed 
B  selfed 
BXA 
BX20 

(BXA)Fisibs 
14  crossed 
14  crossed 
16  selfed 
16  crossed 
16  crossed 
18  selfed 
20X16 
20XA 
16Xcousin 
1 8  crossed 
18  crossed 
20  selfed 
20X16 
20  crossed 
20  crossed 
22  selfed 
22  crossed 
22  crossed 
24  selfed 
24  crossed 
24  crossed 
30  selfed 
28  crossed 
28  crossed 
22  crossed 
24  crossed 

21 
29 
18 
8 
30 

io 

7 
3 
4 
5 
1 
3 
66 

51 
70 
25 
39 
55 

8 
53 
32 
23 
22 
50 
31 
29 

5 

30 
6 

12 

3 

21 

44 

32 

55 

17 

15 

35 

18 

20 

3 

4 

44 

18 

21 

74 

71 

57 

28 

25 

10 

58 

6 

26 

13 

14 

25 

2 

4 

2 

1 

16 

3 

5 

ii 

1 
4 
1 

4 

1 

'46 
50 

9 
33 
32 

5 
11 
16 
17 
18 

9 
20 
40 
23 
18 
51 
11 
19 
14 

4 
46 

9 
18 
18 
14 
12 

'3 

4 

102 

105 
55 
50 

106 
63 
96 
98 
44 
41 
94 
50 
53 
74 
96 

102 
31 
60 

115 
86 

108 
51 
51 
40 
86 
80 
84 
48 
41 
81 
35 
48 
73 
96 
85 
36 
47 
51 
87 

113 
52 
67 
91 
32 
44 
97 
34 
21 
68 
39 
44 

100 

104 

18 
20 
21 
22 
20 
24 
17 
19 
24 
24 
18 
19 
19 
17 
24 
26 
24 
26 
28 
27 
25 
25 
23 
26 
28 
28 
27 
28 
29 
24 
25 
26 
20 
27 
24 
23 
28 
28 
24 
28 
30 
29 
26 
25 
27 
23 
27 
27 
25 
29 
28 
29 
29 

24.0 

24.8 

21.0'i 

22.5/ 

44-8 

35.3 

17.5 

25.0 

17. 3\ 

18. 3J 

23,5 

9.5 
10,3 

9.8 
54  0 
60.0 
16,3 

29  8 
61,3 
50.3 
41  0 
29,5 

30  0 
7.3 

49.8 

49  0 

41,5 

23,81 

20  8/ 

21.0 

22.51 

24.0/ 

17.3 

53.8 

46  0 

11.3 

26.01 

28.5/ 

20.3 

63  3 

29, 0\ 

34  8/ 

25,3 

17,51 

26.8/ 

28  0 

14  81 

12.8/ 

11  5 

14,31 

19,5/ 

37,3 

53 , 5 

53  0 
33.7 

59.2 

60,3 
80  0 

D4.6 

D4.7 

DS1.81 

D52.82.  .  .  . 

D6.9 

D7.10i.... 
D7.102.  ..  . 

D8.11 

D8  1' 

45 
7 
2 

5 
7 

■2 

"3 
12 

38 

15 

10 

8 

4 

9 

19 

42 

27 

22 

19 

17 

23 

36 

54 

2 

5 

4 

5 

6 

5 

9 

3 

15 

"8 

6 

15 

43 

'  '4 

5 

4 

27 

36 

12 

19 

30 

12 

5 

22 

2 

'  '4 

18 

"1 

1 

1 

8 

9 

23 

21 

41 

11 

8 

36 

10 

2 

2 

5 

4 

27 

28 

"3 

"i 

8 

14 

12 

4 

10 

16 

7 

5 

1 

5 

4 

11 

21 

"5 
7 
4 

"s 

12 
11 
3 
4 
3 
6 
2 
7 

'  '2 
1 

'2 
5 
3 
7 

12 
5 
7 

'4 

'    1 
1 
1 
2 

14 
8 
8 
1 

'2 

"2 
9 

5 
5 

"1 

li 
1 
3 

4 

■  ■ 

4 
'2 

26.0 
56.4 

59.7 

35.7 
27.1 
27.6 
18  8 
80  4 

D8.13 

D8.14i.... 
D8  142 

::: 

1 
2 

84.0 
74.9 
70.8 

D8.15 

D8.16 

D9.17 

DIO1.I81.  . 
DIO2.I82.  . 
D 1 1  19 

1 

2 
3 
2 

1 
8 

32 

5 
5 

76.2 
83.5 
54,2 

83^3' 
25  9 

1)11.20.... 
Dll  M 

19 

82  6 

87,5 

D13.22.... 

D14i.23i. 

D142.232 

1 
2 

70,6 
71.4 

D15.24.... 
D16i.25i 

1 

37.0 
80.0 

D16'.252 

D17.26.... 

33  8 

D17  27 

80  0 

D17.28.... 
D17.29.    . 

77.3 
44.6 

DI81  3O1 

79.4 

DI81.3O2 

D19.31 

33  3 

D19.32.... 

80.0 

D20i.33i 

76.5 

D2O2.332 

D21.34.. 

39.6 

D22  35i 

83.2 

D22.352 

1 

D23.36.... 

41.3 

D24  37i 

71.4 

D24.372 

D'5  39 

24.2 

D26.40i 

3 

2 

23 

18 

4 

2 

31 

22 

58.1 

1)26.402 

1)27  41 

5 
3 

81.8 

D28  42 

73.5 

Totals 

214 

570 

846 

588 

497 

341 

261 

123 

73 

48 

36 

24 

16 

5 

6 

3 

4 

S6S'; 

Without- 

Se)jfftrti)«* 

tjon. 

live  Greneyatiew'X 


1 1  k  I  i 


f7  libtr-u. 


I 

■' 

I 

1 

^L 

■t.. 

wm^ 

Bi. 

Fig.  2.4 — An  exhibit  set  up  in  the  Genetics  Department  of  Cornell  University  in  1910,  dis- 
playing materials  grown  at  the  Station  for  Experimental  Evolution  in  1909. 


Fig.  2.5— The  best  eleven  ears  of  the  highest-yielding  selfed  line  (F  29.70  in  Table  2.7) 
grown  in  1911  {top  row) ;  the  best  eleven  ears  of  the  best  Fi  hybrid  grown  in  the  same  year 
(F  32.75  in  Table  2.7) ;  and  the  best  eleven  ears  of  a  crossbred  strain  (F  55.84  in  Table  2.7) 
in  which  selling  was  completely  prevented  during  five  years.  This  shows  the  relative  vari- 
ability which  is  characteristic  of  these  three  types  of  families,  the  Fi  being  no  more  variable 
than  the  inbred,  while  the  crossbred  is  quite  noticeably  more  variable. 


34 


GEORGE  HARRISON  SHULL 


summary  form.  Some  73  ears  were  selected  for  planting,  and  5,343  ears  were 
harvested.  The  complete  grain-row  distribution  was  as  follows: 

Grain-rows 8       10       12       14       16     18     20     22     24     26     28     30     32     34     36     38     40     42   Total 

Frequencies ...  .   401     812  1271     921     716476275   141   118     74     53     41     24       8       6       4       1       1     5343 
Percentages.  ...   7.5   15.2  23.8  17.2  13.4  89  5.2  2.6  2.2   14  10  0.8  0.5  0.2  0.1  0.1  00  0.0  100.0 


To  save  space  and  still  indicate  as  completely  as  possible  the  significant 
results  of  these  studies  in  1910,  the  data  from  the  several  kinds  of  families 
of  white  dent  corn  grown  at  the  Station  for  Experimental  Evolution  that 
year  are  presented  in  the  form  of  averages  in  Table  2.6.  The  several  quanti- 
tative indicators  of  physiological  vigor,  namely,  the  average  number  of 
grain-rows,  heights  of  stalks,  and  bushels  of  ear-corn  per  acre,  can  be  readily 
compared  as  follows: 


Types  of  Families 


Inbreds  selfed .... 
InbredsXsibs.  .  .  . 

Crossbreds 

Fi  between  inbreds 
F2  from  Fi  selfed. . 
F2  from  Fi  X  sibs .  . 


No.  of 

Av.  No.  of 

Families 

Grain-Rows 

10 

12.6 

8 

13.7 

11 

16.9 

6 

15.2 

11 

13.3 

11 

13.5 

Av.  Heights 
in  Dms. 


19.3 
19.8 

23.5 
25.7 
23.3 
23.1 


Av.  Yields 
in  Bu./A. 


25.0 
28.7 
63.5 
71.4 
42.6 
47.9 


Si.x  interesting  comparisons  can  be  made  among  these  summaries:  (1) 
comparisons  between  inbreds  selfed  and  inbreds  crossed  with  pollen  from 
one  or  more  of  their  sibs;  (2)  comparisons  between  inbreds  and  crossbreds 
in  which  selfing  has  been  completely  prevented,  but  which  still  represent  a 
(fairly  low)  degree  of  inbreeding;  (3)  comparisons  between  inbreds  and  their 
Fi  hybrids;  (4)  comparisons  between  the  crossbreds  in  which  selfing  has  been 
prevented  through  six  generations  and  the  Fi  hybrids  in  which  five  successive 
generations  of  selfing  have  been  succeeded  by  a  single  cross;  (5)  comparisons 
between  the  Fi  and  the  F2  hybrids  of  the  inbreds;  and  (6)  comparisons  be- 
tween F2  hybrid  families  produced  by  selfing  the  Fi  and  those  F2  families 
produced  by  sibcrosses  in  the  Fi. 

On  making  these  comparisons  we  see  that  the  evidence  for  residual  hetero- 
zygosity in  the  inbreds  is  indicated  by  excesses  in  the  sibcrossed  families  of 
the  inbreds  over  the  selfed  inbreds  of  8.7  per  cent  in  grain-row  number,  2.8 
per  cent  in  heights  of  stalks,  and  14.7  per  cent  in  yield  of  ear-corn.  In  the  Fo 
families  (sections  E  and  F,  of  Table  2.6)  those  produced  from  sibcrosses  in 
the  Fi  surpass  those  families  produced  from  selfings  in  the  Fi  by  0.9  per  cent 
in  grain- row  number  and  12.5  per  cent  in  yield. 

The  average  heights  of  stalks  reverse  the  expectation  by  showing  an  in- 
significantly less  height  from  the  sibcrossed  matings  than  from  the  selfings, 
the  difference  being  0.9  per  cent.  The  contrast  between  the  results  of  six 
successive  selfings  and  the  continued  prevention  of  selfing  for  the  same  six 


TABLE  2.6 

AVERAGE  V.ALUES  IN  THE  FAMILIES  OF  WHITE  DENT  MAIZE 

GROWN  IN  1910,  GROUPED  ACCORDING  TO  THE 

TYPES  OF  MATING  OF  THE  PARENTS 


Pedigree 
Numbers 

Parental 
Grain-Rows 

Number 
of  Stalks 

Av.  No. 
of  Grain- 
Rows 

Heights 
in  Dms. 

Wts.  in 
Lbs.  Av. 

Yields 
Bu./A. 

(A)   Families  from  Inbreds  Selfed 

El. 16  .  . 

8  selfed 
8  selfed 
10  selfed 
12  selfed 
A(8)  selfed 
B(14)  selfed 
14  selfed 
18  selfed 
22  selfed 

26.  28  selfed 

■ 

57 
83 
79 
80 
75 
53 
66 
82 
62 
72 

10.0 

9.0 
11.1 
12.3 

8.8 
12.9 
13.8 
15.2 
17.9 
15.2 

17 

18 

20 

17 

16.5 

24 

23 

19 

19 

19 

9.8 
22.0 
18.3 
11.4 
9.1 
7.3 
16.3 
15.3 
11.0 
17.5 

24  4 

E2.19 

^9  6 

E7.29 

33  9 

E9.32 

20  9 

E11.34 

E19.47 

E24.54 

E26.56 

E34.67 

E36.71 

18.1 
11.0 
25.8 
22.9 
19.2 
34.2 

Unweighted  averages 

71 

12.6 

19.3 

10.7 

25.0 

(B)   Families  from  Inbreds  Pollinated  by  Sibs;  Selfing  Prevented 

El. 17 

lOXsibs 
10  X  sib 
12  X  sib 
A(8)Xsib 
B(12)Xsib 
18  X  sib 
20  X  sib 
?(fasc.)Xsib 

61 

75 
85 
55 
54 
89 
65 
73 

10.2 
9.9 
11.0 
9.5 
12.7 
15.8 
17.9 
22.5 

19 
18 
22 
16 
24 
20 
20 
20 

13.8 
21.0 
18.3 
7.5 
5.3 
24.5 
15.3 
18.3 

29  8 

E2.20 

39  5 

E7.30 

37  ^ 

E11.33 

E19.48 

E26.57 

E34.68 

E36.72 

16.0 
7.8 
37.8 
25.6 
35.2 

Unweighted  averages 

61 

13.7 

19.8 

15.5 

28.7 

(C)  Families  from  Parents  Given  Mi.^ed  Pollen  in  Each  Generation; 
Selfing  Prevented 

E3.23    ... 

8,  10  crossbred 
10  crossed 
12  crossed 
14  crossed 
16  crossed 
18  crossed 
20  crossed 
20,  22  crossed 
24,  20  crossed 
32  crossed 
32  crossed 

88 
65 
91 
94 
95 
202 
100 
45 
69 
56 
99 

9.5 
10.3 
13.2 
13.7 
14.9 
16.0 
18.5 
20.0 
24.2 
19.2 
26.2 

22 

22 

24 

27 

28 

22.5 

23 

21 

22 

24 

23 

30.8 
31.0 
51.0 
49.0 
48.8 
76.8 
35.8 
26.3 
24.5 
22.5 
39.0 

49  9 

E8.31 

68  1 

E18.46 

E23.53 

E25.55 

E30.63 

E33.66 

E35.70 

E37.73 

E40.75 

E40.76 

80.1 
74.5 
73.3 
54.3 
51.1 
83.3 
50.7 
57.4 
56.3 

Unweighted  averages 

91.3 

16.9 

23.5 

39.6 

• 

63.5 

TABLE  2.6 — Continued 


Pedigree 

Numbers 

Parental 
Grain -Rows 

Number 
of  Stalks 

Av.  No. 
of  Grain- 
Rows 

Heights 
in  Dms. 

Wts.  in 
Lbs.  Av. 

Yields 
Bu./A. 

(D)  Fi  Hybrids  between  Different  Inbred  Lines 

E2.21 

A(10)X16 

A(10)XB 

A(8)X10 

A(8)XB 

18X14 

18X26  + (fasc.) 

95 
94 
95 
84 
109 
92 

13.8 
12.8 
11.0 
12.3 
17.8 
23.3 

24 
28 

25 
25 
27 
25 

50.3 
50.0 
33.5 
28.5 
60.8 
62.5 

75.6 

E2.22    

76.0 

E11.36 

E11.37 

E26.58 

E34.69 

51.5 
48.5 
79.6 
97.1 

Unweighted  averages 

93 

15.2 

25.7 

47.6 

71.4 

(E)  F2  Families  from  FiXSelf 

E4.24 

(lOXA)Fi  selfed 
(10Xl4)Fiselfed 
(A  X20)Fi  selfed 
(AX 22)Fi  selfed 
(AX  16)Fi  selfed 
(A  XB)Fi  selfed 
(BXA)Fi  selfed 
(BX20)Fi  selfed 
(20Xl6)Fi  selfed 
(20XA)Fi  selfed 
(20  Xl6)Fi  selfed 

86 
86 
76 
83 
94 
96 
95 
92 
97 
95 
93 

10.6 
12.1 
13.9 
12.8 
12.8 
12,0 
11.7 
15.1 
16.6 
13.0 
15.9 

21 

22 
19.5 

24 
25 
25 
24 

25 
25 
22 
24 

30.8 
29.8 
20.5 
18.8 
33.5 
24.0 
25.3 
28.0 
35.3 
22.0 
29.5 

51.1 

E5.26   

49.4 

E12.38 

E13.40 

E15.42 

E16.44 

E20.49 

E21.51 

E27.59 

E28.61 

E32.64 

38.5 
31.4 
50.9 
35.7 
38.0 
43.5 
51.9 
33.1 
45.3 

Unweighted  averages 

90.3 

13.3 

23.3 

27.0 

42.6 

(F)   F2  Families  from  FiXSibs 

E4.25 

(10Xl2)F,Xsibs 

(10Xl4)FiXsibs 

(AX20)FiXsibs 

(AX22)FiXsibs 

(AXl6)FiXsibs 

(AXB)FiXsibs 

(BXA)FiXsibs 

(BX20)FiXsibs 

(20Xl6)FiXsibs 

(20XA)FiXsibs 

(20Xl6)FiXsibs 

85 
83 
80 
96 
95 
93 
80 
93 
89 
92 
97 

10.7 
12.2 
14.2 
13.4 
12.3 
11.8 
11.6 
15.5 
17.2 
13.7 
15.4 

21 
22 
21 
25 
23 
24 
24 
25 
25 
23 
21 

31.3 
35.0 
28.8 
27.0 
37.3 
21.0 
23.5 
31.8 
37.3 
30.0 
25.3 

52.5 

E5.27 

60.2 

E12.39 

E13.41 

E15.43 

E16.45 

E20.50 

E21.52 

E27.60 

E28.62 

E32.65 

51.3 
40.2 
56.0 
32.3 
42.0 
48.8 
59.8 
46.6 
37.6 

Unweighted  averages 

89.4 

13.5 

23.1 

29.8 

47.9 

BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  37 

years  (sections  A  and  C,  Table  2.6)  shows  the  latter  in  excess  of  the  former 
by  34.0  per  cent  in  grain-row  number,  22.1  per  cent  in  height  of  stalks,  and 
154.2  per  cent  in  per  acre  yields  of  ears.  The  superiority  of  the  Fi  hybrids 
between  different  inbreds  and  the  families  in  which  selfing  had  been  pre- 
vented during  six  generations  of  controlled  breeding  (sections  D  and  C, 
Table  2.6),  is  indicated  by  an  excess  in  heights  of  stalks  of  the  Fi  families 
over  the  crossbreds,  of  9.4  per  cent,  and  in  yields  of  ear-corn  per  acre  of  12.3 
per  cent.  But  here  there  is  a  notable  reversal  in  grain-row  numbers.  Not- 
withstanding these  proofs  of  the  superior  vigor  of  the  F/s  over  the  cross- 
breds, the  latter  exceed  the  former  in  grain-row  number  by  10.8  per  cent. 

The  reason  for  this  reversal  is  easily  recognized  when  we  consider  that 
parents  were  selected  in  these  studies  for  their  grain-row  numbers,  with  no 
noticeable  selection  for  heights  and  yields.  In  section  D  of  Table  2.6,  we  note 
that  only  one  parent  of  any  of  the  Fi  families  had  a  grain-row  number  in 
excess  of  18.  The  crossbred  families  ranged  in  parental  grain-row  numbers 
from  8  to  32.  Five  of  the  families  came  from  parents  having  more  than  18 
rows  of  grains. 

To  make  a  fair  comparison  between  the  two  types  of  breeding  in  their  re- 
lation to  grain-row  number,  it  is  necessary  to  use  only  the  crossbred  families 
having  parents  with  no  more  than  18  grain-rows.  When  we  make  such  a  limi- 
tation, we  find  the  average  grain-row  number  for  the  remaining  six  crossbred 
families  is  only  12.9.  The  grain-row  average  for  the  six  Fi  families,  namely, 
15.2,  exceeds  the  crossbreds  by  17.1  per  cent.  Limiting  the  other  indicators 
of  physiological  vigor  to  the  same  six  crossbred  families,  we  find  that  the  F/s 
exceed  the  corresponding  crossbreds  on  the  average  by  6.3  per  cent  in  height 
of  stalks  and  7.0  per  cent  in  yield  of  ear-corn. 

In  1911  I  was  again  in  full  personal  charge  of  the  corn  experiments  at 
the  Station  for  Experimental  Evolution,  and  was  able  to  expand  the  work 
considerably,  both  quantitatively  and  in  the  types  of  matings  studied. 
We  planted  84  cultures  in  the  white  dent  series  as  well  as  25  cultures  of 
other  types  of  corn.  The  total  number  of  white  dent  ears  of  which  the  grain- 
rows  were  counted  was  6,508  which  showed  the  following  frequencies: 

Grain-rows 8       10       12       14       16       18     20     22     24     26     28     30     32     34     36  Total 

Frequencies 267     767   1725  1298     931     683  363  164  114     95     65     23       7       3       3     6508 

Percentages 4.1   11.8  26.5   19.9  14.3  10.5  5.6  25  1.8  1.5  0.9  0.4  0.1  0.1  0.1     99  9 

In  Table  2.7  the  1911  results  are  presented  in  condensed  form.  Families 
are  grouped  in  eleven  sections  representing  fairly  homogeneous  groups, 
mostly  based  on  the  types  of  matings  involved.  Sections  D  and  E  are  both 
made  up  of  the  same  five  families  of  F2  hybrids  produced  by  selfing  the  same 
number  of  different  Fi's.  For  these  families  each  seed  ear  was  used  to  plant 
two  rows.  The  one  row  of  each  such  family  was  grown  with  the  other  cultures, 
as  usual,  in  the  East  Garden.  The  second  row  of  each  of  these  families  was 


TABLE  2.7 

AVERAGE  GRAIN-ROW  NUMBERS  AND  YIELDS  PER  ACRE  OF  WHITE 

DENT  MAIZE  GROWN  IN  1911  GROUPED  ACCORDING  TO  THE 

TYPES  OF  MATINGS  OF  THE  PARENTS 


Pedigree 
Numbers 

Parental  Strains  Involved 

Number 
of  Stalks 

Av.  Num- 
ber Grain- 
Rows 

Weights 
in  Lbs. 

Yields 
Bu./A. 

(A)  Families  from  Inbreds  Selfed 

FI6.681 

E2.68.2 

F29.70 

F32.73 

F34.76 

F0.77 

E19.79i 

F47.792 

F0.80 

E24.82 

F56.85 

E36.92 

F74.94 

8  selfed 

8  selfed 

10  selfed 

12  selfed 

Strain  A  selfed 

A  from  L.  H.  Smith 

B  selfed 

B  selfed 

B  from  L.  H.  Smith 

16  selfed 

20  selfed 

26,  28  selfed 

*"Cobs"  selfed 

12 
44 
89 
95 
98 
101 
3 
46 
95 
84 
90 
79 
64 

8.7             1.5           17.9 

9.0             6.0           19.5 

10.9            16.5            26.5 

11.8            11.3            16.9 

8.4             8.3            12.1 

8.9             8.8            12.4 

Not  counted  nor  weighed 

Not  counted  nor  weighed 

14.3              4.3              6.8 

14.0              7.5            12.8 

15.3            13.8           21.8 

22.7            11.5            20.8 

Not  counted  nor  weighed 

1 

Unweighted  averages  (omit- 
ting the  three  uncounted 
families) 

78.7 

12.4 

8.9 

16.7 

(B)   Families  from  Parents  Given  Mixed  Pollen  in  Each 
Generation;  Selfing  Prevented 

F23.69 

F31.72 

F46.78 

F53.81 

F55.84 

F632.86 

F66.87 

F70i.91 

F73.93 

F76.96 

8  crossed 
10  crossed 
12  crossed 
14  crossed 
16  crossed 
18  crossed 
20  crossed 
22  crossed 
24  crossed 
32  crossed 

71 
95 
92 
97 
101 
105 
99 
63 
68 
94 

10.4 
10.7 
12.2 
13.7 
15.2 
18.2 
19.4 
22.3 
23.8 
25.2 

30.3 
30.3 
44.5 
40.8 
33.0 
42.5 
40.0 
20.8 
34.5 
50.5 

60.2 

45.5 
69.1 
60.0 

46.7 
51.8 
57.7 
45.9 
72.5 
60.4 

Unweighted  averages 

88.5 

17.0 

36.7 

57.0 

(C)  Fi  Hybrids  between  Different  Inbreds 

F29.71 

F32.74 

F32.75 

F54.83 

(10Xl2)Fi 
(lOXB)Fi 
(10Xl6)Fi 
(16X20)Fi 

62 
106 
100 
100 

12.2 
12.8 
14.3 
18.4 

24.5 
65.3 
63.0 
58.2 

56.5 
87.9 
90.0 
83.2 

Unweighted  averages 

92 

14.4 

52.7 

79.4 

♦  This  was  a  slightly  fasciated  brevistylis  type,  with  silks  about  half  as  long  as  the  husks.  Usually  it  pro- 
duced no  grains  except  when  given  artificial  help. 


38 


TABLE  2.7— Contimie 


Pedigree 

Number 

Parental  Strains  Involved 

Number 
of  Stalks 

Av.  Num- 
ber Grain- 
Rows 

Weights 
in  Lbs. 

Yields 
Bu./A. 

(D)   F2  Families  from  Fi  Selfed,  Clrown  in  Annex  No.  1 

F21.24 

F22.28 

F36.31 

F37.36 

F58.54 

(8X20)F,  selfed 
(8XB)F,  selfed 
(A XlO)Fi  selfed 
(A  XB)Fi  selfed 
(20  XI 6) Fi  selfed 

69 
61 
99 
93 
103 

13.8 
13.4 
11.3 
11.8 
16.2 

23.0 
31.3 
33.3 
17.0 

54.3 

47.6 
73.2 
48.0 
29.3 

47.5 

Unweighted  averages 

83 

13.3 

31.8 

49.1 

(E)    Same  Families  as  in  (D),  but  Grown  in  East  Garden 

F21.24 

F22.28 

F36.31 

F37.36 

F58.54 

(8  X20)Fi  selfed 
(8XB)Fi  selfed 
(AX  10)F,  selfed 
(A  XB)F,  selfed 
(20  Xl6)Fi  selfed 

98 

101 

98 

76 

97 

13.4 
13.4 
11.1 
11.0 
16.8 

36.0 
56.0 
31.3 
15.3 
34.3 

52.5 
79.2 
45.9 
28.7 
50.8 

Unweighted  averages 

94 

13.2 

34.6 

51.4 

(F)   F2  Families  from  FiXsibs,  All  Grown  in  East  Garden 

F21.25 

F22.29 

F36.34 

F37.37 

F58.55 

(8X20)FiXsib 

(8XB)FiXsib 

(AXlO)FiXsibs 

(AXB)F,Xsib 

(20Xl6)FiXsib 

59 
97 
93 
71 
110 

12.9 
12.8 
10.8 
11.3 
16.0 

22.0 
42.8 
26.3 
18.5 
35.0 

53.3 
63.0 
40.3 
37.2 
45.5 

Unweighted  averages 

86 

12.8 

28.9 

47.9 

(G)   Fs  Families  from  F2  Selfed 

F38.39 

F40.42 

F42.45 

F44.46 

F49.49 

F51.52 

F59.57t 

F59.57 

F61.59 

F64.62 

(A  X20)F.,  selfed 
(AX22)F.2  selfed 
(A Xl6)Fo  selfed 
(A  XB)F..  selfed 
(16XA)F2  selfed 
(16  X20)Fo  selfed 
(20 Xl6)Fo  selfed 
(20X16)F2  selfed 
(20  XA)F2  selfed 
(BX16)F.2  selfed 

84 
108 

67 

92 
112 

95 
100 
100 
117 
107 

13.0 
11.6 
10.2 
11.0 
11.4 
15.0 
15.9 
16.4 
12.0 
17.0 

9.8 
19.3 

10.5 

6.0 

24.3 

23.8 

24.5 

25.5 

9.8 

12.5 

16.6 
25.5 
22.4 
9.3 
30.9 
35.7 
35.0 
36.4 
13.6 
16.7 

Unweighted  averages 

98.2 

13.3 

16.6 

24.2 

t  This  family  was  divided  and  this  section  was  grown  in  the  North  Hill-field.  All  of  the  other  families  were 
grown,  as  usual,  in  East  Garden. 


39 


TABLE  2.1— Continued 


Pedigree 
Numbers 

Parental  Strains  Involved 

Number 
of  Stalks 

Av.  Num- 
ber Grain- 
Rows 

Weights 
in  Lbs. 

Yields 
Bu./A. 

(H)  Fs  Families  from  FiXSibs 

F38.40 

F40.43 

F44.47 

F49.50 

F59.58 

F61.60 

F64.63 

(AX20)F2Xsib 

(AX22)F2Xsib 

(AXB)F,Xsib 

(16XA)F2Xsib 

(20Xl6)F.2Xsib 

(20XA)F,Xsib 

(BXl6)F2Xsib 

106 
112 

94 
104 

90 
111 
104 

13.5 
11.9 
11.2 
11.8 
16.5 
13.8 
15.1 

26.0 
26.5 
21.8 
29.8 
38.5 
25.0 
27.5 

35.0 
33.8 
33.1 
40.9 
61.1 
32.2 
37.8 

Unweighted  averages 

103 

13.4 

27.9 

39.1 

(I)   Families  from  "Three-Way"  and  Iterative  Crosses 

F58.56 

F74.95 

F21.27 

F22.30 

F36.33 

F27.38 

F51.53 

(20Xl6)FiX22 

"Cobs"X(20Xl6)Fi 

(8X20)FiX20 

(8XB)F,XB 

(AXB)FiXA 

(AXB)F,XB 

(16X20)F2X20 

114 
29 
67 

103 
84 
79 

108 

18.9 
20.6 
15.0 
14.3 
10.5 
12.8 
17.1 

61.8 
23.3 
28.5 
37.8 
23.0 
23.5 
42.3 

77.4 
114.5 
60.8 
52.4 
39.1 
29.8 
55.9 

Unweighted  averages  (three- 
way) 

71.5 

19.7 

42.5 

96.0 

Unweighted  averagesj  (iter- 
ative) 

83.3 

13.1 

28.2 

45.5 

(K)    Families  from  "Four-Way"  Crosses,  the  So-called  "Double-Cross"' 

F21.26 

F36.35 

F69.66 

F36  32§ 

(8X20)F,X(AXlO)Fi 
(AX10)F,X(20X16)F, 
(22X"Cobs")F,X(8XlO)Fi 
(AXlO)FiX(AXB)Fi 

67 
106 

75 
102 

12.7 
12.8 
16.3 
11.2 

28.5 
47.0 
58.5 
45.5 

60.8 

63.3 

111.4 

63.7 

Unweighted  averages 

87.5 

14.3 

44.9 

74.8 

(L)   F3  Families  from  Four-Way  F2  Crosses,  and  Imperfect 
Iteratives  of  Same  Form 

F61.61 

F38.41 

F40.44 

F44.48 

F49.51 

(20XA)F2X(BX16)F2 

(AX20)F..X(AX22)F2 

(AX22)F2X(AX16)F2 

(AXB)F2X(16XA)F2 

(16XA)F2X(16X20)F2 

102 
103 
110 

78 
117 

15.3 
12.9 
13.2 
11.4 
13.3 

31.8 
27.0 
43.5 
28.0 
44.3 

44.5 
37.5 
56.5 
51.3 
61.6 

Unweighted  averages 

102 

13.2 

34.9 

50.3 

X  Does  not  include  F51.53. 

§F36.32  is  an  imperfect  4-way,  being  partly  iterative,  involving  only  3  inbreds. 


40 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT  41 

planted  in  new  plots  of  ground  about  one-fourth  mile  north  of  the  original 
Station  grounds. 

The  purpose  of  this  replication  was  to  determine  the  degree  of  consistency 
of  results  secured  in  these  new  locations  with  those  recorded  for  the  cultures 
grown  in  the  different  conditions  of  soil,  drainage,  exposure,  lighting,  etc.,  in 
the  East  Garden.  Summaries  of  these  two  sections  of  Table  2.7  show  the  cul- 
tures grown  in  the  new  plot  with  average  grain-row  number  1.29  per  cent 
higher  than  in  the  same  families  grown  in  the  East  Garden.  However,  the 
East  Garden  cultures  produced  a  higher  average  yield  of  ear-corn  by  4.70 
per  cent. 

Comparison  between  selfing  and  sibcrossing  was  made  a  subject  of  special 
study  in  the  inbred  and  Fi  families  in  1910.  This  was  not  continued  in  1911 
in  the  inbreds,  but  was  given  a  further  test  in  the  derivation  of  the  F2  families 
from  the  Fi,  and  was  carried  forward  to  the  derivation  of  F3  families  from  the 
F2.  These  comparisons  as  they  relate  to  Fi  families  are  given  in  sections  E 
and  F  of  Table  2.7.  They  show  the  F2  families  derived  from  selfing  their  Fi 
parents  slightly  superior  to  those  Fo  families  produced  from  sibcrosses  in 
the  Fi.  This  is  indicated  by  an  average  grain-row  number  3.1  per  cent  higher 
and  average  yield  7.5  per  cent  higher  in  the  Fo  families  from  selfed  Fi  par- 
ents, thus  reversing  the  indications  from  the  1910  cultures. 

The  comparison  of  selfing  versus  sibcrossing  in  the  production  of  the  F3  by 
these  two  methods  of  breeding  in  F2  can  be  derived  from  section  G  for  selfings 
and  section  H  for  the  sibcrosses.  Summaries  of  these  two  sections  show  a 
superiority  from  sibcrosses  of  0.4  per  cent  in  average  grain-row  number  and 
61.6  per  cent  in  yield.  A  part  of  this  discrepancy  is  clearly  due  to  the  inclu- 
sion of  families  in  the  selfed  group  which  had  no  direct  counterpart  in  the 
sibcrossed  group.  If  we  limit  the  comparison  to  the  families  which  are  repre- 
sented in  both  groups,  we  can  avoid  this  cause  of  distortion.  We  then  find 
the  sibcrossed  families  superior  to  the  selfed  by  1.5  per  cent  in  grain-row 
number,  and  48.6  per  cent  in  yields. 

Comparative  values  between  inbreds  and  crossbreds,  as  shown  in  sections 
A  and  B  of  Table  2.7,  and  between  crossbreds  and  Fi  hybrids,  are  essentially 
the  same  as  in  1910.  The  ratios  of  inbreds,  crossbreds,  and  Fi  hybrids,  with 
respect  to  yields,  is  0.29  to  1.00  to  1.22.  Again  the  average  grain-row  number 
is  less  in  the  Fi  than  in  the  crossbreds,  and  for  the  same  reason.  This  particu- 
lar group  of  Fi  families  came  from  parents  with  low  average  grain-row  num- 
bers, as  compared  with  the  broader  parentage  of  the  crossbreds. 

The  relationship  of  F.-s  to  Fo  can  now  be  noted  by  comparing  the  results 
in  sections  G  and  H  of  Table  2.7,  with  sections  D,  E,  and  F.  There  are  sev- 
eral ways  in  which  such  comparisons  can  be  made.  Perhaps  as  good  a  way 
as  any  is  simply  to  combine  all  of  the  F2's  together,  regardless  of  the  con- 
siderations which  led  these  to  be  tabulated  in  three  separate  sections,  and 
compare  the  results  with  all  the  F3  families  of  sections  G  and  H  likewise 


42  GEORGE  HARRISON  SHULL 

averaged  in  an  undivided  population.  When  treated  in  this  way,  we  find  that 
the  Fo's  have  an  average  grain-row  number  of  13.1  and  average  yields  of  49.5 
bushels  per  acre,  while  the  F3  had  an  average  of  13.4  grain-rows  and  pro- 
duced an  average  of  30.4  bushels  per  acre.  If  we  associate  the  average  yield 
of  the  Fi  families,  79.0  with  these  values  for  Fo  and  F3,  we  see  the  beginning 
of  the  characteristic  curve  in  which  the  loss  of  yield  from  one  generation  to 
the  next  is  about  twice  as  great  as  the  loss  for  the  next  following  generation . 
It  remains  to  consider  the  last  three  sections  of  Table  2.7,  in  which  are 


Fig.  2.6 — Total  yields  of  ear  corn  of  two  selfed  strains,  Strain  16  and  Strain  20,  in  the  fore- 
ground (exaggerated,  of  course,  by  foreshortening),  and  their  Fi,  F2,  and  F3  hybrids,  left  to 
right,  successively,  in  the  background.  As  may  be  seen  in  Table  2.7,  these  yields,  calculated 
in  terms  of  bushels  per  acre,  are  12.76  and  21.82  for  the  two  inbreds,  and  83.21,  50.81, 

and  36.43  for  the  three  hybrid  families. 

included  the  results  of  more  complicated  crossing  which  had  become  possible 
through  the  accumulation  of  simpler  crossing  in  preceding  years.  In  section  I 
are  given  two  "three-way"  crosses  and  four  iterative  crosses  involving  Fi 
combinations  and  one  iterative  cross  involving  an  F2  combination,  each  repre- 
senting a  cross  between  a  hybrid  and  an  inbred.  As  might  be  e.xpected,  these 
seven  families  although  similar  in  form  show  no  special  consistency,  since 
they  involve  various  combinations  of  five  different  inbreds  and  five  different 
hybrids. 

In  Table  2.7,  section  K,  are  presented  what  I  believe  to  be  the  first  "four- 
way"  or  so-called  "double  crosses"  ever  made  among  inbreds.  The  elements 
of  one  of  these  double  crosses  are  shown  in  Figure  2.7.  These  double  crosses 
were  made  some  five  or  six  years  before  Dr.  D.  F.  Jones  pointed  out  the 
potentialities  of  such  crosses  in  producing  hybridized  seed  corn  at  a  price 


X 


X 


10 


(Ax  10)  F, 


Fig.  2.7 — One  of  the  first  four-zcay  or  double  crosses  ever  grown  from  selfed  strains  of  maize. 
The  single  crosses  for  this  double  cross  were  made  in  1909,  the  cross  between  the  Fi's  was 
made  in  1910,  and  the  double-cross  ear  at  bottom  (G35.62)  was  grown  in  1911  and  grains 

from  it  were  used  for  i)lanting  in  1912. 


44  GEORGE  HARRISON  SHULL 

that  could  make  the  pure-line  method  of  corn  production  practical.  No  credit 
is  sought  for  the  fact  that  I  made  these  four-way  crosses  some  years  prior  to 
the  similar  combinations  made  by  Dr.  Jones.  They  are  presented  here  only 
because  they  belong  in  a  historical  account. 

In  the  last  section  of  Table  2.7  I  have  entered  five  families  which  have  the 
form  of  four-way  crosses,  but  in  which  the  single  crossings  used  were  Fo  in- 
stead of  Fi.  Only  the  first  of  these  five  families  actually  involved  four  differ- 
ent inbreds,  the  others  being  partially  iterative,  in  that  only  three  inbreds 
contributed  to  each.  A  comparison  of  the  double  crosses  both  of  Fi  and  F2, 
with  the  corresponding  single  crosses,  is  instructive.  Comparison  of  the  sum- 
mary of  section  C  with  that  of  section  K  shows  the  double  cross  families 
slightly  inferior  to  the  single  cross  families,  as  indicated  by  a  1  per  cent  higher 
grain- row  number  and  6  per  cent  higher  yield  of  the  single  cross  families 
over  the  double  cross.  Comparing  sections  L  and  E,  it  is  to  be  noted  that  the 
double  cross  retains  the  vigor  of  the  F2,  instead  of  declining  to  the  vigor  of 
the  F3  families  produced  by  the  usual  methods,  as  seen  in  sections  G  and  H, 
Table  2.7. 

In  1911 1  realized  that  the  effective  exposition  of  the  important  discoveries 
we  were  making  required  photographs  of  prepared  exhibits.  A  number  of  such 
exhibits  were  set  up  and  photographed,  and  have  been  presented  in  lantern 
slides  on  many  occasions.  I  have  included  the  most  instructive  of  these  here. 

Here  the  detailed  account  of  these  studies  must  end,  for  although  they 
were  continued  in  1912, 1  have  been  unable  to  locate  the  field  and  harvesting 
notes  including  grain-row  counts  and  weighings  for  the  1912  cultures.  These 
1912  cultures  were  especially  designed  to  explore  the  evidences  of  Mendelian 
segregations  in  the  F2  and  the  F3  families,  with  respect  to  grain-row  num- 
bers and  yields.  They  included  11  families  of  the  breeding  Fi  X  self,  8  families 
of  Fi  X  sib,  21  F2  X  self,  10  Fo  X  sibs,  and  five  families  of  F3  X  self.  There 
was  also  an  interesting  pair  of  approximations  to  eight-way  combinations  or 
quadruple  crosses  produced  by  reciprocal  combinations  of  the  four-way 
crosses  included  in  the  1911  cultures.  While  these  had  the  form  of  quadruple 
crosses,  they  were  imperfect  in  that  one  of  the  inbreds  was  repeated,  so  that 
only  seven  different  inbreds  were  represented,  instead  of  eight.  This  was  in- 
evitable since  I  initiated  only  seven  inbred  lines  in  the  beginning  of  these 
experiments. 

The  1912  crop  completed  the  experimental  work  with  corn  at  the  Station 
for  Experimental  Evolution,  and  I  spent  the  next  year  in  Berlin,  Germany. 
In  a  lecture  I  gave  at  Gottingen  about  three  weeks  before  the  beginning  of 
the  first  World  War  the  word  heterosis  was  first  proposed.  I  used  the  occasion 
to  discuss  the  bearing  of  the  results  of  these  studies  on  the  practical  work  of 
breeders  of  various  classes  of  organisms,  both  plant  and  animal.  I  stressed 
the  point  that  the  breeder  should  not  be  content,  as  had  long  been  the  case, 
to  seek  merely  to  avoid  the  deterioration  incident  to  inbreeding,  but  should 


10 


18 


10 


13 


1904  1905  1906  1507  1908  1909  I'JiO  igu  191.5 


SELECTION  FOR  ROW-NIJHBER 
IN   CROSS-FERTILIZED  MAIZE 


"■?58=^ 


^;2^"- 


»-^. 


;o>*\ 


SELECTIOtJ  FOR  RaJ-WIKDER 
IN  SELF-FEHTILIZEU  MIZE 


1904 


1905 


1906 


1907 


1908 


1909 


1910 


1911 


1912 


Fig.  2.8 — Diagrams  of  the  progressive  results  of  selection  for  grain-row  number  under  the 
two  systems  of  breeding:  selling  completely  prevented  in  the  upper  diagram;  selling  the 
sole  method  of  breeding  in  the  lower.  The  numbers  on  the  lines  indicate  the  numbers  of 
rows  of  grains  on  the  parent  ears.  The  circles  show  by  their  position  on  the  scale  at  left  the 
average  grain-row  numbers  of  the  resulting  [progenies. 


46 


GEORGE  HARRISON  SHULL 


recognize  in  heterosis  a  potent  source  of  practical  gains,  to  be  investigated, 
understood,  and  utilized  as  a  new  tool  in  deriving  from  plant  and  animal 
life  their  maximum  contributions  in  the  service  of  man. 

Although  no  further  experimental  work  was  done  with  corn  at  the  Station 
for  Experimental  Evolution  after  1912,  I  tried  to  resume  the  work  in  my 
first  two  years  at  Princeton  University,  by  planting  77  cultures  of  pedigreed 

8 


V,     7 

u 

■;  i 

& 

^3^ 

«» 

V  * 

f 

V- 

♦  •  < 

3 

*" 

z 

^H*-^' 


V, 


'^  /'         %fi>  X  ^h    QA  >S^ 


A 


^^^ 


3  /^  X  B 


S    3. 


Fig.  2.9 — Ears  of  my  white  dent  "strain"  of  corn  grown  at  Princeton  University  in  1916. 
The  ears,  each  typical  of  the  progeny  to  which  it  belonged,  are  from  left  to  right :  SA,  Shull's 
Strain  A;  SA  X  BA,  Fi  hybrid  between  Shull's  Strain  A  and  Biakeslee's  "branch"  of  the 
same  strain;  BA  X  SA,  reciprocal  of  the  last;  BA,  Shull's  Strain  A,  after  two  successive 
sellings  by  Dr.  A.  F.  Blakeslee;  B.\  X  B,  Fi  between  Biakeslee's  branch  of  Strain  A  and 
Shull's  Strain  B;  and  SB,  Shull's  Strain  B.  About  as  much  heterosis  is  shown  by  a  cross  be- 
tween two  sub-hnes  of  Strain  A  as  between  one  of  these  sub-lines  and  Strain  B,  the  impli- 
cation being  that  something  more  specific  may  be  involved  in  this  example  of  heterosis  than 
the  mere  number  of  genetic  difTerences.  (Photo  by  W.  Ralph  Singleton  in  1945.) 

corn  in  1916  and  65  in  1917. 1  used  some  of  the  materials  from  these  cultures 
for  laboratory  studies  in  biometry  in  my  classes  in  genetics.  The  interesting 
results  shown  in  Figure  2.9  are  from  my  1916  crop  at  Princeton.  The  plantings 
at  Princeton  were  made  late  and  the  young  plants  were  decimated  by  pigeons 
and  crows,  so  that  some  valuable  connections  were  lost,  and  with  them  some 
of  my  interest  in  their  continuation. 

As  we  all  know,  heterosis  is  not  limited  to  corn,  and  my  own  interest  in 
the  matter  was  in  no  wise  restricted  to  its  manifestation  in  corn.  There  were 
examples  presented  in  many  other  of  my  genetical  experiments.  I  was  par- 
ticularly interested  in  the  discovery  of  such  special  mechanisms  as  balanced 
lethal  genes  in  the  Oenotheras  and  self-sterility  genes  in  Capsella  grandiflora 


BEGINNINGS  OF  THE  HETEROSIS  CONCEPT 


47 


which,  along  with  many  types  of  asexual  reproduction  including  partheno- 
genesis, specifically  enable  the  organisms  possessing  these  special  mecha- 
nisms to  maintain  the  full  advantages  of  heterosis.  On  one  occasion,  one  of 
my  new  hybrid  combinations  in  Oenothera  happened  to  be  planted  through 
an  area  in  my  experimental  field  where  the  soil  had  become  so  impoverished 
that  none  of  my  other  cultures  reached  their  normal  growth.  Many  of  the 


Fig.  2.10 — The  Fi  hybrids  between  a  cultivated  form  of  Helianthus  anniius  and  a  wild  form 
of  the  same  species  received  from  Kansas.  This  photograph,  taken  at  the  Station  for  Ex])eri- 
mental  Evolution  in  1906,  shows  the  author  alfixing  a  glassine  hag  to  a  head  of  one  of  the 
hybrid  plants.  The  two  parents  of  this  hybrid  averaged  from  5  to  6  feet  tall,  while  51  of 
these  Fi  hybrids,  measured  on  August  28,  1906,  ranged  in  height  from  6.7  to  14.25  feet,  the 
average  being  10.46  feet.  This  may  be  considered  my  iirst  experience  with  hybrid  vigor. 


48  GEORGE  HARRISON  SHULL 

plants  remained  rosettes  or  formed  only  weak  depauperate  stems.  But  this  new 
hybrid  became  a  vigorous  upstanding  form  in  this  impoverished  area  as  well 
as  on  better  soil  elsewhere.  I  recorded  this  as  a  notable  example  of  making 
heterosis  take  the  place  of  manure  or  commercial  fertilizers. 

Figure  2.10  is  a  notable  hybrid,  which  represents  my  first  direct  personal 
contact  with  a  recognized  case  of  hybrid  vigor.  This  hybrid  resulted  from  a 
cross  I  made  in  1905  between  the  so-called  "Russian"  sunflower  and  the  wild 
Helianthus  annuus  of  our  western  prairies.  Both  of  these  forms  have  been  re- 
ferred, botanically,  to  the  same  species.  Both  are  of  approximately  equal 
height,  scarcely  as  tall  as  the  six-foot  step-ladder  shown  in  the  figure.  The 
tallest  of  these  Fi  hybrids  was  14.25  feet  in  height. 

Returning  now  to  the  question  which  I  sidestepped  in  the  beginning — 
what  we  mean  by  the  expression  the  heterosis  concept — I  suggest  that  it  is  the 
interpretation  of  increased  vigor,  size,  fruitfulness,  speed  of  development, 
resistance  to  disease  and  to  insect  pests,  or  to  climatic  rigors  of  any  kind, 
manifested  by  crossbred  organisms  as  compared  with  corresponding  inbreds, 
as  the  specific  results  of  unlikeness  in  the  constitutions  of  the  uniting  parental 
gametes. 

I  think  the  first  clear  approach  to  this  concept  was  involved  in  a  statement 
which  I  have  already  quoted,  that  "a  different  explanation  was  forced  upon 
me"  (in  my  comparisons  of  cross-fertilized  and  self-fertilized  strains  of 
maize).  That  is,  "that  self-fertilization  simply  serves  to  purify  the  strains, 
and  that  my  comparisons  are  not  properly  between  cross-  and  self-fertiliza- 
tion, but  between  pure  strains  and  their  hybrids."  Since  heterosis  is  recog- 
nized as  the  result  of  the  interaction  of  unlike  gametes,  it  is  closely  related  to 
the  well  known  cases  of  complementary  genes.  It  differs  from  such  comple- 
mentary genes,  however,  mainly  in  being  a  more  "diffuse"  phenomenon  in- 
capable of  analysis  into  the  interactions  of  specific  individual  genes,  even 
though  it  may  conceivably  consist  in  whole  or  in  part  of  such  individual 
gene  interactions. 


H.  K.  HAYES 

University  of  Minnesofa 


Chapter  3 

Development  of 
the  Heterosis  Concept 


Hybrid  vigor  in  artificial  plant  hybrids  was  first  studied  by  Koelreuter  in 
1763  (East  and  Hayes,  1912).  The  rediscovery  of  Mendel's  Laws  in  1900 
focused  the  attention  of  the  biological  world  on  problems  of  heredity  and  led 
to  renewed  interest  in  hybrid  vigor  as  one  phase  of  quantitative  inheritance. 

Today  it  is  accepted  that  the  characters  of  plants,  animals,  and  human 
beings  are  the  result  of  the  action,  reaction,  and  interaction  of  countless 
numbers  of  genes.  What  is  inherited,  however,  is  not  the  character  but  the 
manner  of  reaction  under  conditions  of  environment.  At  this  time,  when 
variability  is  being  expressed  as  genetic  plus  environmental  variance,  one 
may  say  that  genetic  variance  is  the  expression  of  variability  due  to  geno- 
typic  causes.  It  is  that  part  of  the  total  variance  that  remains  after  eliminat- 
ing environmental  variance,  as  estimated  from  studying  the  variances  of 
homozygous  lines  and  Fi  crosses  between  them. 

Early  in  the  present  century.  East,  at  the  Connecticut  Agricultural  Ex- 
periment Station,  and  G.  H.  Shull  at  Cold  Spring  Harbor,  started  their 
studies  of  the  effects  of  cross-  and  self-fertilization  in  maize.  The  writer  has 
first-hand  knowledge  of  East's  work  in  this  field  as  he  became  East's  assist- 
ant in  July,  1909,  and  continued  to  work  with  him  through  1914.  In  1909, 
East  stated  that  studies  of  the  effects  of  self-  and  cross-pollination  in  maize 
were  started  with  the  view  that  this  type  of  hiformation  was  essential  to  a 
sound  method  of  maize  breeding.  In  addition  to  studies  of  maize,  which  is 
normally  cross-pollinated,  East  carried  out  studies  in  tobacco  of  crosses  be- 
tween varieties  and  species.  This  gave  an  opportunity  of  studying  the  effects 
of  self-  and  cross-pollination  with  a  self-pollinated  jilant.  A  1912  j)ublication 
of  East  and  Hayes  made  the  following  statement: 

The  decrease  in  vigor  due  to  inbreeding  naturally  cross-fertilized  species  and  the  increase 
in  vigor  due  to  crossing  naturally  self-fertilized  species  are  manifestations  of  one  phenome- 
non. This  phenomenon  is  heterozygosis.  Crossing  produces  heterozygosis  in  all  characters 
by  which  the  parent  plants  differ.  Inbreeding  tends  to  produce  homozygosis  automatically. 

49 


50  H.  K.  HAYES 

Several  photographs  from  this  bulletin  are  of  some  interest.  A  picture  of 
two  inbred  lines  of  maize  and  their  Fi  cross  was  one  of  the  first  published  field 
views  of  hybrid  vigor  from  crossing  inbred  lines  of  maize.  East  told  me  that 
such  a  demonstration  of  hybrid  vigor  would  create  a  sensation  if  the  material 
had  been  grown  in  the  corn  belt. 

Some  Fi  crosses  between  species  and  sub-species  in  tobacco  gave  large  in- 
creases in  vigor.  Some  species  crosses  were  sterile.  Some  varietal  crosses 
within  species  showed  little  or  no  increase  in  vigor,  other  crosses  gave  an  aver- 
age increase  of  25  per  cent  in  height  over  the  average  of  their  parents.  A  few 
wide  species  crosses  were  very  low  in  vigor.  One  such  cross  beween  Nicotiana 
tabacum  and  Nicotiana  alata  graiidiflora  was  sterile  and  very  weak  in  growth. 
Photographs  of  the  parents  and  hybrids  bring  out  the  fact  that  a  lack  of  vigor 
in  a  few  cases  was  known  to  accompany  the  heterozygous  condition.  Natural- 
ly such  undesirable  combinations  had  little  importance  either  to  the  plant 
breeder  or  as  a  basis  for  evolution. 

In  1910,  G.  H.  ShuU  summarized  the  effects  of  inbreeding  and  crossbreed- 
ing in  maize  in  a  clear,  concise,  and  definite  manner.  The  student  of  heredity 
in  this  early  period  had  little  conception  of  the  complexity  of  inheritance. 
Hybrid  vigor  was  in  many  cases  not  clearly  Mendelian.  The  term  heterosis 
was  coined  by  Shull  and  first  proposed  in  1914.  He  used  the  term  to  avoid 
the  implication  that  hybrid  vigor  was  entirely  Mendelian  in  nature  and  to 
furnish  a  convenient  term  to  take  the  place  of  such  phrases  as  "the  stimulus 
of  heterozygosis." 

At  this  time  it  was  usually  stated  that  increased  vigor  in  hybrids  was  due 
to  a  more  rapid  cell  division  as  stimulated  by  the  heterozygous  condition  of 
the  genotype.  A.  F.  Shull  in  1912  attributed  the  vigor  "to  the  effect  of  a 
changed  nucleus  and  a  (relatively)  unaltered  cytoplasm  upon  each  other." 

The  purpose  of  this  chapter  is  to  discuss  some  phases  of  the  development 
of  the  heterosis  concept  since  1910.  Three  main  topics  will  be  presented  cover- 
ing utilization,  breeding  methods,  and  genetic  concepts  with  particular  ref- 
erence to  practical  applications  and  to  genetic  explanations. 

UTILIZATION  OF  HETEROSIS  BY  THE  PRODUCER 

The  presentation  of  East  and  Hayes  in  1912  emphasized  the  probable 
practical  value  of  heterozygosis.  A  review  of  experiments  with  maize  was 
made.  In  discussing  Shull's  (1909)  plan  for  the  use  of  single  crosses  between 
inbred  lines,  it  was  stated  that  the  procedure  was  desirable  in  theory  but 
difficult  of  application.  At  this  early  time  the  inbred  lines  of  maize  that  were 
available  seemed  so  lacking  in  vigor  that  the  use  of  Fi  crosses  between  selfed 
lines  in  maize  for  the  commercial  crop  seemed  impractical.  Both  Shull  and 
East  believed  that  some  method  of  direct  utilization  of  hybrid  vigor  in  maize 
would  be  found. 

One  is  inclined  to  forget  that  the  inbred  lines  of  maize  of  today  are  marked- 


DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT  51 

ly  superior,  on  the  average,  to  those  of  1910.  Jones's  discovery  about  1917 
of  the  double  cross  plan  of  producing  hybrid  seed  in  maize,  and  the  subse- 
quent proof  by  many  workers  that  double  crosses  can  be  obtained  that  closely 
approach  the  vigor  of  Fi  crosses  between  selfed  lines,  furnished  the  basis  for 
the  utilization  of  hybrid  vigor  in  field  corn.  With  sweet  corn,  however,  F] 
crosses  between  selfed  lines  are  used  very  widely  today  for  the  commercial 
crop. 

East  and  Hayes  emphasized  that  Fi  crosses  probably  would  be  of  com- 
mercial value  in  some  truck  crops  where  crossing  was  easy.  Eggplants,  to- 
matoes, pumpkins,  and  squashes  were  considered  to  offer  promise  for  a  prac- 
tical use  of  such  vigor.  The  writers  also  mentioned  the  fact  that  heterozygosis 
had  been  used  in  vegetatively  propagated  plants,  though  not  purposely,  and 
that  it  seemed  feasible  to  make  a  practical  application  in  the  field  of  forestry. 

The  use  of  heterosis  in  practical  plant  and  animal  improvement  has  borne 
out  and  surpassed  these  early  predictions  as  shown  in  Table  3.1. 

TABLE  3.1 

USE  OF  HETEROSIS  IN  CROP  PLANTS  AND  LIVESTOCK 

Farm  crops:  Maize,  sugar  beets,  sorghums,  forage  crops,  and  grasses 

Horticultural  crops:  Tomatoes,    squashes,    cucumbers,     eggplants,     onions, 

annual  ornamentals 
Silkworms 

Livestock:  Swine,  poultry,  beef  and  milk  cattle 

Vegetatively  propa- 
gated plants 

In  the  corn  belt  of  the  United  States  nearly  100  per  cent  of  all  maize  is 
hybrid.  Hybrid  corn  is  rapidly  being  developed  in  other  countries  of  the 
world,  and  is  one  of  the  best  illustrations  of  the  practical  utilization  of  mod- 
ern genetics.  Considerable  evidence  leads  to  the  conclusion  that  heterosis  can 
be  used  extensively  in  farm  crops,  including  such  widely  different  plants  as 
sugar  beets,  sorghums,  tobacco,  forage  crops,  and  grasses. 

With  horticultural  plants,  where  the  individual  plant  is  of  rather  great 
value,  planned  heterosis  has  proven  worth  while.  First  generation  crosses 
of  tomatoes,  onions,  egg  plants,  cucumbers,  and  squashes  have  proven  their 
value  and  are  being  grown  extensively  by  home  and  truck  gardeners.  Similar 
use  is  being  made  of  heterosis  in  some  annual  ornamentals. 

Heterosis  has  become  an  important  tool  of  the  animal  breeder.  Its  use  in 
silkworm  breeding  is  well  known.  Practical  utilization  of  hybrid  vigor  has 
been  made  in  swine  and  poultry,  and  applications  are  being  studied  with  beef 
cattle,  dairy  cattle,  and  sheep.  A  somewhat  better  understanding  of  the 
effects  of  inbreeding  and  crossing  by  the  breeder  has  aided  in  applications 
with  livestock.  As  in  plants,  inbreeding  makes  controlled  selection  possible, 
while  controlled  crosses  may  be  grown  to  utilize  favorable  gene  combinations. 


52  H.  K.  HAYES 

METHODS  OF  BREEDING  FOR  HETEROSIS 

In  general  there  is  a  much  closer  relation  between  the  characters  of  par- 
ents and  of  their  Fi  crosses  in  self-pollinated  plants  than  between  the  char- 
acters of  inbred  lines  of  cross-pollinated  plants  and  their  Fi  crosses. 

Characters  of  Parents  and  Fi  Crosses  in  Self-pollinated  Plants 
A  recent  study  by  Carnahan  (1947)  in  flax,  which  is  normally  self-polli- 
nated, may  be  used  for  illustrative  purposes.  Four  varieties  of  flax  were  se- 
lected to  represent  desirable  parental  varieties.  Each  was  crossed  with  four 
other  varieties,  of  different  genetic  origin  from  the  first  group,  to  be  used  as 
testers.  Sufficient  seed  for  Fi  and  F2  progenies  was  produced  so  that  all 

TABLE  3.2 

PARENT  AND  Fi  CROSSES,  YIELD 
IN  BUSHELS  PER  ACRE* 

Parent  Tester  Varieties 

Varieties  5  6  7  8 

16  14  17  13 


1 

19 

31 

25 

22 

19 

2 

18 

24 

26 

19 

20 

3 

13 

26 

24 

20 

18 

4 

17 

22 

21 

20 

19 

*  Parent  yields  outside   rectangle,   Fi   crosses 
within. 

progenies  could  be  planted  in  replicated,  8-foot  rows  at  the  rate  of  200  seeds 
per  row.  Combining  ability  was  studied  in  Fi  and  F2  in  comparison  with  the 
parents  for  yield  of  seed,  number  of  seeds  per  boll,  number  of  bolls  per  plant, 
weight  of  1000  seeds,  date  of  full  bloom,  and  plant  height. 

As  shown  in  Table  3.2,  each  Fi  cross  yielded  more  than  its  highest  yielding 
parent,  although  for  one  cross  the  difference  was  only  slightly  in  favor  of 
the  Fi.  For  an  average  of  all  crosses,  the  Fi  yielded  40  per  cent  more  than  the 
average  of  the  parents,  and  the  Fo,  26  per  cent  more.  The  lowest  yielding 
cross,  3X8,  was  produced  from  a  cross  of  the  two  lowest  yielding  parents. 
The  highest  yielding  cross,  1X5,  however,  could  have  been  selected  only 
by  actual  trial.  It  was  obtained  by  crossing  the  highest  yielding  selected 
variety  with  the  second  highest  yielding  tester  variety. 

There  was  excellent  agreement,  on  the  average,  for  each  of  the  characters 
studied  between  the  average  expression  of  the  characters  of  the  parents  and 
their  Fi  crosses.  Carnahan  concluded  that  for  each  character  studied  there 
appeared  to  be  a  good  relationship  between  the  performance  of  the  parents 
and  the  average  performance  of  their  Fi  crosses.  The  characters  of  the  par- 
ents in  this  study  were  as  good  or  better  indication  of  the  combining  ability 
of  a  parental  variety  as  that  obtained  from  a  study  of  average  combining 
ability  in  four  crosses. 


DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT 


53 


Powers  (1945)  obtained  also  relatively  good  agreement  in  tomatoes  be- 
tween the  parental  yield  of  10  varieties  and  that  of  all  possible  Fi  crosses 
between  the  10  varieties  (see  Table  3.3). 

Moore  and  Currence  (1950)  in  tomatoes  made  a  somewhat  comparable 
study  to  that  of  Carnahan  with  flax.  They  used  two  three-way  crosses  as 
testers  for  a  preliminary  evaluation  of  combining  ability  of  27  varieties. 
Based  on  this,  eight  varieties  were  selected  that  gave  a  wide  range  in  aver- 
age combining  ability  for  sev^eral  characters  including  early  yield  and  total 
yield.  These  varieties  were  crossed  in  all  combinations,  and  yield  trials  of  the 

TABLE  3J 

YIELD  OF  RIPE  FRUIT  IN  GRAMS 
IN  TOMATOES  (AFTER  POWERS) 


Yield  of  Ripe  Fruits  (per  Plant) 

Variety  or  Inbred 

Variety  or  Inbred 

9  Crosses  (av.) 

Grams 

Grams 

L.  escidsntum 

Bounty  4101 

513+   39 

1280  +  53 

4102 

607+  86 

1267  +  46 

4105 

332+   64 

1081+33 

4106 

828  +  108 

1236  ±45 

Es.XL.  pirn 

4103 

1066+159 

1597  +  54 

4104 

808  +  114 

1340  +  44 

4107 

801  +  111 

1181+47 

4108 

857  + 108 

1192+41 

4109 

1364+151 

1968  +  46 

4110 

1868  +  149 

2231  +  52 

varieties  and  Fi  crosses  were  made.  There  was  relatively  good  agreement 
between  the  early  test  for  combining  ability  and  the  average  yield  of  Fi 
crosses,  but  the  relationship  did  not  seem  superior  to  the  varietal  performance 
as  a  means  of  predicting  combining  ability  in  crosses.  In  the  studies  by  Carna- 
han, Moore,  Currence,  and  Powers  the  only  means  of  selecting  the  most  de- 
sirable Fi  cross  was  by  actual  trial. 

Characters  of  inbred  Lines  and  Their  Fi  Crosses  in  Maize 

Numerous  studies  have  been  made  with  maize  of  the  relation  between 
characters  of  inbred  lines  and  of  their  Fi  crosses.  There  usually  have  been 
indications  of  significant  correlations  for  most  characters  of  inbred  lines  and 
their  Fi  crosses.  In  most  cases,  however,  the  relationship  was  not  very  large 
or  highly  important  when  one  studied  individual  characters,  or  the  more  com- 
plex character — yield  of  grain.  The  studies  have  been  reviewed  by  numerous 
workers  (see  Sprague,  1946b). 


54  H.  K.  HAYES 

Hayes  and  Johnson  (1939)  in  Minnesota  studied  the  relation  between  the 
characters  of  110  inbred  lines  of  maize  and  their  performance  in  top  crosses. 
The  characters  studied  in  selfed  lines  in  replicated  yield  trials  are  given  in 
Table  3.4. 

All  possible  correlations  were  made  between  the  individual  characters  of 
the  inbreds  and  of  these  characters  and  the  yield  of  grain  of  top  crosses.  The 

TABLE  3.4 

CHARACTERS  OF   110  INBRED  LINES  IN 
CORN  CORRELATED  WITH  INBRED- 
VARIETY  YIELDING  ABILITY 

1.  Date  silked  7.  Stalk  diameter 

2.  Plant  height  8.  Total  brace  roots 

3.  Ear  height  9.  Tassel  index 

4.  Leaf  area  10.  Pollen  yield 

5.  Pulling  resistance  11.   Grain  yield 

6.  Root  volume  12.  Ear  length 

TABLE  3.5 

TOT.\L  CORRELATIONS  BETWEEN  CHARACTERS  OF  110  INBREDS, 
LABELED  1  TO  12,  AND  YIELDING  ABILITY  OF  INBRED- 
VARIETY  CROSSES  DESIGNATED  AS  15 


Characters  Correlated 

2             3             4             5 

6 

7 

8 

9 

10 

11 

12 

15 

1 

0.51     0.61     0.48     0.65 

0.62 

0.55 

0.38 

0.37 

0.22 

0.07 

-0.06 

0,47 

2 

0.76     0.44    0.48 

0.43 

0.40 

0.26 

0.19 

0,36 

0.25 

0.08 

0.27 

3 

0.43     0.54 

0.50 

0.41 

0.35 

0.33 

0.22 

0.15 

-0.01 

0.41 

4 

0.50 

0.44 

0.48 

0.40 

0.29 

0.18 

0.20 

0.08 

0.29 

5 

0.76 

0.51 

0.60 

0.41 

0.21 

0.15 

0.04 

0.45 

6 

0.55 

0.74 

0.39 

0.29 

0.19 

0.03 

0.54 

7 

0.54 

0.24 

0.27 

0.21 

0.15 

0.41 

8 

Multiple  value  of  R 

0.26 

0.22 

0.20 

0.07 

0.45 

9 

for  inbred-variety  yield 

0.20 

-0.00 

0.03 

0.19 

10 

and  twelve  characters  of 

0.35 

0.32 

0.26 

11 

inbred  =  0.67 

0.64 

0.25 

12 

0.28 

Significant  value  of  r  for  P  of  .05  =  0.19. 
Significant  value  of  ;■  for  P  of  .01  =  0.25. 

characters,  in  general,  were  those  that  were  considered  to  evaluate  the  in- 
breds in  developmental  vigor. 

The  total  correlations  between  characters  are  summarized  in  Table  3.5. 
Most  correlations  were  significant  at  the  5  per  cent  or  1  per  cent  point  ex- 
cept the  relation  between  ear  length  and  other  characters  of  the  inbreds.  All 
relationships  between  the  characters  of  the  inbreds,  including  grain  yield,  and 
the  yield  of  top  crosses  were  significant  at  the  1  per  cent  point  except  for 
tassel  index  of  the  inbreds,  and  that  was  significant  at  the  5  per  cent  point. 
The  multiple  correlation  coefficient  of  0.67  indicated  that  under  the  condi- 
tions of  the  experiment  about  45  per  cent  of  the  variability  of  inbred-variety 


DEVELOPMENT  OF  THE   HETEROSIS  CONCEPT 


55 


yield  was  directly  related  to  characters  of  the  inbreds.  These  relationships 
between  the  parents  and  their  Fi  crosses  were  somewhat  larger  than  those 
obtained  by  others  with  maize.  Nevertheless,  relationships  were  much 
smaller  than  has  been  obtained  in  similar  studies  with  self-pollinated  plants. 

Richey  (1945b)  compared  the  yield  of  inbred  parents  in  the  S3  and  S4  gener- 
ations of  selling  with  the  mean  yield  of  their  single  crosses  from  data  taken 
by  Jenkins  and  Brunson.  Similar  comparisons  were  made  between  the  yield 
in  top  crosses  and  the  mean  yield  in  single  crosses  (see  Table  3.6). 

Although  for  various  reasons  the  r  values  are  not  strictly  comparable,  the 
yield  of  inbreds  was  as  strongly  correlated  with  the  mean  yield  of  their 
single  crosses  as  the  yield  in  top  crosses  was  correlated  with  the  mean  yield 
of  single  crosses. 

TABLE  3.6 

CORRELATION  COEFFICIENTS  FOR  YIELDS  OF 

INBRED  PARENTS  OR  TOP  CROSSES  WITH 

MEAN  YIELDS  OF  SINGLE  CROSSES* 


Hybrids 
Correlated 

Previous  Generations 
OF  Inbreeding 

WITH 

S3t 

84 

Inbred  parents 
Top  crosses 

.25,  .64,  .67 
.53 

.41,  .45 
.53 

*  After  Richey,  after  Jenkins  and  Brunson. 
t  S3  =  three  years  selfed,  etc. 


Comparison   of  Methods  with   Self-  and   Cross-pollinated   Plants 

In  self-pollinated  plants  it  seems  probable  that  the  first  natural  step  in 
the  utilization  of  heterosis  normally  may  consist  of  the  selection  of  available 
parental  varieties  that  in  themselves  produce  the  best  combination  of  char- 
acters. It  seems  important  to  continue  breeding  for  the  best  combination  of 
genes  that  can  be  obtained  in  relatively  homozygous  varieties.  Where  hybrid 
seed  can  be  produced  cheaply  enough,  or  new  methods  can  be  found  to 
make  crosses  more  easily,  heterosis  can  be  used  to  obtain  from  the  hybrid  an 
advance  in  productivity  over  the  homozygous  condition. 

In  cross-pollinated  plants  two  general  methods  of  breeding  for  heterosis 
are  now  being  widely  utilized.  One  consists,  as  in  maize,  of  the  selection  with- 
in and  between  selfed  lines  and  the  use  of  single,  three-way,  or  double  crosses 
for  the  commercial  crop.  The  second  general  method  consists  of  selecting 
or  breeding  desirable  clones  of  perennial  crops.  These  are  evaluated  for  com- 
bining ability  by  polycross,  or  other  similar  methods,  and  the  desirable  clones 
used  to  produce  Fi  crosses,  double  crosses,  or  synthetic  varieties. 


56  H.   K.   HAYES 

There  seems  to  be  some  difference  of  opinion  regarding  the  selection  proc- 
ess in  its  application  to  maize  improvement.  One  school  of  thought  practices 
a  somewhat  similar  method  of  breeding  selfed  lines  as  is  used  in  self-pollinated 
plants,  with  the  viewpoint  that  controlled  selection  makes  it  possible  to  iso- 
late in  the  inbred  lines  the  genes  for  characters  needed  in  the  hybrids.  Ap- 
parently the  relationship  between  the  characters  of  inbreds  and  their  Fi 
crosses  will  become  greater  as  inbred  lines  themselves  improve.  The  other 
extreme  of  viewpoint  (Hull,  1945a)  is  that  the  greater  part  of  hybrid  vigor  is 
due  to  interallelic  interaction  of  genes  to  such  an  extent  that  selection  based 
on  appearance  may  be  harmful.  In  a  recurrent  selection  program  Hull, 
therefore,  does  not  recommend  selection  for  vigor  of  growth,  although  he 
states  that  plants  showing  pest  or  weather  damage  should  be  avoided. 

It  is  probable  that  differences  between  these  two  so-called  schools  may 
have  been  overstated.  Both  believe  that  the  actual  test  for  combining  ability 
in  hybrid  combination  is  necessary.  The  stage  in  the  breeding  program  when 
such  test  should  be  made  will  depend  on  the  material  worked  with  and  the 
nature  of  the  breeding  program.  In  both  cross-  and  self-pollinated  plants  an 
actual  trial  will  be  needed  to  determine  the  combination  that  excels  in 
heterosis. 

Where  clonal  lines  can  be  propagated  vegetatively,  a  method  of  selecting 
for  heterosis  in  alfalfa  was  suggested  by  Tysdal,  Kiesselbach,  and  Westover 
(1942),  by  means  of  polycross  trials.  The  method  is  being  used  extensively 
today  with  perennial  forage  crops  that  normally  are  cross-pollinated.  The 
writer  is  studying  the  method  with  early  generation  selfed  lines  of  rye.  With 
perennial  crop  plants,  selection  for  combining  ability  is  made  for  heterozy- 
gous parent  clones.  Where  disease  and  insect  resistance  or  winter  hardiness 
are  important,  it  may  be  essential  to  insure  that  the  clones  used  in  the  poly- 
cross trials  excel  for  these  characters.  Polycross  seed  is  produced  on  selected 
clones  under  open-pollinated  conditions  where  the  clones  are  planted  together 
at  random  under  isolation. 

In  one  study  of  progenies  of  eight  clones  by  Tysdal  and  Crandall  (1948) 
yields  were  determined  from  polycross  seed  in  comparison  with  top  cross  seed 
when  each  of  the  clones  was  planted  in  isolation  with  Arizona  common  alfalfa 
(see  Table  3.7).  The  agreement  for  combining  ability  was  relatively  good  in 
the  two  trials. 

An  early  suggestion  of  utilization  of  heterosis  in  alfalfa  was  by  double 
crosses,  from  single  crosses  between  vegetatively  propagated  clones,  without 
entire  control  of  cross-pollination.  Synthetic  varieties  also  have  been  sug- 
gested as  a  means  of  the  partial  utilization  of  heterosis.  In  one  comparison 
the  progeny  of  a  synthetic  combination  of  four  clones  of  high  combining 
ability  yielded  11  per  cent  more  forage  than  a  similar  combination  of  four 
clones  of  low  yielding  ability.  A  recent  comparison  of  eight  synthetics  led 
Tysdal  and  Crandall  to  conclude  that  the  first  synthetic  and  second  syn- 


DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT 


57 


thetic  seed  progenies  gave  about  the  same  forage  yield.  In  this  comparison, 
heterosis  continued  through  the  second  seed  increase  of  the  high  yielding 
synthetic. 

Other  Studies  with  Maize 

Combining  ability,  that  is  ability  to  yield  in  hybrid  combination,  has  been 
shown  by  various  workers  to  be  an  inherited  character  (Hayes  and  Johnson, 
1939),  (Cowan,  1943),  (Green,  1948).  It  seems  feasible  to  breed  for  high  com- 
bining ability  as  for  other  quantitative  characters.  In  the  breeding  program 

TABLE  3.7 

FORAGE  YIELDS  OF  POLY- 
CROSSES  COMPARED  TO 
TOP  CROSSES  OF  THE 
SAME  CLONES* 


Yield  Relative  to 

Grimm 

AS    100 

Clone  No. 

Arizona 

Polycross 

Top  Cross 

1 

121 

130 

2 

111 

122 

3 

101 

117 

4 

'J9 

103 

5 

97 

105 

6 

96 

101 

7 

89 

101 

8 

76 

101 

*  After  Tvsdal  and  Crandall. 


for  the  production  of  imj^roved  inbred  lines,  it  is  often  possible  to  select  as 
parents  of  crosses,  select  lines  having  high  combining  ability  as  parents  of 
crosses,  in  addition  to  selection  for  other  characters  that  are  desired.  In 
breeding  for  heterosis,  however,  it  seems  evident  that  genetic  diversity  of 
parentage  is  equally  as  important  as  combining  ability  (see  Hayes,  and 
Immer,  1942;  Sprague,  1946b). 

All  relatively  homozygous,  inbred  lines  in  maize  are  much  less  vigorous 
than  the  better  Fi  crosses.  It  is  apparent  that  heterosis  is  of  great  impor- 
tance in  crosses  with  inbred  lines  of  maize. 

Inbred  lines  that  have  undesirable  characters  may  be  easily  imjjroved  by 
the  application  of  any  one  of  several  methods  of  breeding.  The  breeder  may 
select  for  each  problem  the  method  or  methods  that  seem  to  him  most  ap- 
plicable. In  breeding  selfed  lines  the  selection  of  parents  that  have  comple- 
mentary characters  that  together  include  the  characters  desired  in  the  im- 
proved inbred  is  a  natural  first  step.  Subsequent  methods  of  breeding  may 


58 


H.  K.  HAYES 


be  used  according  to  the  viewpoint  of  the  breeder  and  the  particular  prob- 
lem to  be  solved. 

While  combining  ability  is  an  inherited  character,  it  seems  of  special  in- 
terest that  single  crosses  of  high  X  high  combiners  have  not  been  greatly  su- 
perior in  yield,  on  the  average,  to  crosses  of  high  X  low.  Both,  however,  were 
clearly  higher  in  yielding  ability  than  low  X  low  crosses  (Johnson  and 
Hayes,  1940),  (Cowan,  1943),  (Green,  1948).  An  illustration  from  Johnson 
and  Hayes  (Table  3.8)  shows  the  type  of  results  obtained.  The  crosses  were 
classified  for  yielding  ability  in  comparison  with  recommended  double 
crosses  of  similar  maturity. 

Two  recent  studies  in  Minnesota  may  be  used  to  illustrate  other  breeding 
problems.  A  further  study  was  made  by  Johnson  (1950)  of  the  combining 
ability  of  F4  lines  that  were  studied  in  earlier  generations  by  Payne  and 
Hayes  (1949).  Yield  relations  in  the  double  cross  Min.  608  (A344  X  A340) 
(A357  X  A392)  are  illustrated  in  Table  3.9. 


TABLE  3.8 

FREQUENCY  DISTRIBUTION  FOR  YIELD  OF  SINGLE  CROSSES 

OF  SIMILAR  MATURITY  IN  COMPARISON  WITH 

RECOMMENDED  DOUBLE  CROSSES  AS  0 


Type  of 

Class  Centers  of  —1  to  —2,  +1  to  +2,  etc.  Times 
THE  S.E.  OF  A  Difference 

Cross 

-7 
-8 

-5 
-6 

-3 
-4 

-1 

-2 

0 

+  1 
+2 

+  3 
+4 

+  5 
+  6 

+  7 
+  8 

Total 

Mean 

Low  X  low . 

1 
3 
1 

1 

.... 
"5 

2 

11 

12 

4        4 

12 
52 
83 

-0  5  +  0  7 

Low  X  high 

HighXhigh 

1 

6 

8 

16 
35 

9 
20 

5 
4 

1 

+  1.1  +  0.4 
+  1  1+0  2 

TABLE  3.9 

YIELD  RELATIONS  IN  MIN.  608 

( A334  X  A340)  (A357  X  A392) 


%  M. 

Yield 
(Bu.) 

A334XA357  and  A392 

A340XA357  and  A392 

19.6 

18.5 

66.8 
62.4 

Average 

19.0 

64.6 

A357XA334and  A340 

A392XA334and  A340 

19.5 
18.6 

66.0 
63.2 

Average 

19.0 

64.6 

Min.  608 

19.0 

64.0 

DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT 


59 


In  these  studies  the  usual  method  of  predicting  combining  ability  of  a 
double  cross  gave  excellent  agreement  between  both  predictions  and  the 
actual  double  cross  yield. 

The  studies  of  the  performance  in  early  and  later  tests  of  F2  to  F4  lines 
from  L317  X  A116  when  crossed  with  (A334  X  A340)  in  comparison 
with  A357(A334  X  A340)  were  carried  out  by  Payne  and  Johnson.  The 
methods  of  comparing  combining  ability  in  different  generations  were 
adapted  by  the  writer,  who  alone  is  responsible  for  the  conclusions  drawn. 
The  lines  were  first  placed  in  +1,  —  1,  etc.  X  L.S.D.  at  the  5  per  cent  point 
with  the  performance  of  A357(A334  X  A340)  as  0.  Classes  for  performance 
of  individual  lines  were  made  by  adding  the  yield  class  of  a  line  to  its  moisture 
class  with  the  sign  of  the  latter  changed. 

The  F2  and  F3  crosses  were  both  grown  the  same  year,  the  F3  and  F4  were 
grown  in  different  years,  and  the  F4and  the  top  crosses  were  grown  the  same 
year  (see  Tables  3.10,  3.11,  3.12). 

In  these  studies  no  new  lines  seemed  markedly  superior  to  A357  in  com- 


TABLE  3.10 

COMBINING  ABILITY  RELATION  OF  F2  AND  F3  LINES 

OF  (L317XA116)  IN  CROSSES  WITH  (A334XA340) 

GROWN  IN  SAME  TRIAL  IN  1947 


o 
•-t 

o 


+2 

1 

0 

2 

2 

1 

1 

-1 

1 

2 

1 

1 

-2 

1 

2 

3 

-3 

1 

1 

1 

3 

1 

-4 

? 

2 

-5 

2 

-7 

2 

+2     +1     0     -1     -2     -3     -4     -5 
F2  crosses,  performance  classes 


Total 
1 
6 

5 
6 
7 
4 

2 
2 

3i 


TABLE  3.11 

COMBINING  ABILITY  RELATION  OF  F3  AND  F4  LINES 

OF  (L317XA116)  IN  CROSSES  WITH  (A334XA340) 

F3  GROWN  IN  1947,  F4  IN  1949 

Total 


o 

o 

en 


+2 

1 

0 

1 

2 

2 

1 

-1 

5 

-2 

1 

1 

3 

1 

-3 

1     1 

2 

3 

-4 

1 

1 

2 

-5 

1 

1 

-7 

1 

I 

+2     +1     0     -1     -2     -3     -4 
F4  crosses,  performance  classes 


1 
6 

5 
6 
7 
4 
2 
2 

33 


60 


H.  K.  HAYES 


bining  ability  with  (A334  X  A340).  As  A357  is  rather  outstanding  in  com- 
bining ability  the  result  may  not  be  so  surprising.  There  was  much  greater 
relation  between  the  combining  ability  of  F3  and  F4  lines  and  of  F4  with  top 
crosses  than  between  F2  and  F3. 

In  an  unpublished  study  of  gamete  selection,  with  a  different  but  highly 
desirable  double  cross,  there  was  an  indication  that  a  lower  yielding  inbred 
could  be  improved  by  an  application  of  gamete  selection  (Stadler,  1944). 
The  study  is  from  one  phase  of  a  breeding  program  to  improve  Min.  406. 
The  yield  relations  of  inbreds  in  an  average  of  single  crosses  are  given  in 
Table  3.13. 

Approximately  60  Fi  plants  of  A25  X  Golden  King  were  selfed  and  top 
crossed  with  A73  X  A375.  Thirty-two  of  the  more  desirable  plants  were  se- 
lected to  study  in  yield  trials.  In  this  study  both  yield  and  moisture  classes 
of  plus  1,  plus  2,  etc.  X  L.S.D.  at  5  per  cent  were  used  around  the  mean  of 

TABLE  3.12 

COMBINING  ABILITY   RELATION  OF  F4  LINES  OF 

(L317XA116)  IN  CROSSES  WITH  (A334XA340)  AND 

WITH  GOLDEN  KING.  GROWN  IN  1949 

Total 


In 

A334 
X 

A340 
Crosses 


+  2 

1 

+  1 

1 

0 

1     3 

1 

1 

-1 

1 

-2 

9 

1 

3 

-3 

3 

4 

1 

-4 

1 

-5 

1 

1 

+  1     0     -1     -2     -3     -4     -5 
In  Golden  King  Crosses 


1 
1 
6 
1 
13 
8 
1 
2 

33 


TABLE  3.13 

GAMETE  SELECTION  IN  THE  IMPROVE- 
MENT OF  MINHYBRID  406 

(A25XA334)(A73XA375) 


Av.  OF  Crosses 

%M. 

Bu. 

A25XA73,  A375 

A334XA73,  A375 

A73XA25,  A334 

A375XA25,  A334 

24.6 
24.7 

24.6 
24.7 

76.2 
79.4 

74.8 
80.8 

Proposal  for  improvement  of  A25  and  A73: 
A25XG.  King  gametes 
A73XMurdock  gametes 


DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT 


61 


A25  X  tester  as  0.  The  results  (see  Table  3.14)  indicate  that  gametes  from 
Golden  King  are  a  desirable  source  of  improvement  of  A25  in  crosses  with 
A73  X  A375. 

From  this  first  trial  three  high  and  three  low  yielding  lines  were  selected, 
and  selfed  progeny  grown  in  Si.  Plants  in  each  of  the  three  Si  high  and  three 
low  combining  lines  were  selected,  selfed,  and  again  top  crossed  on  A73  X 
A3 75.  The  agreement  for  So  and  Si  lines  was  very  good  (see  Table  3.15). 
It  appears  that  gamete  selection  is  an  excellent  breeding  method  for  the 
early  selection  of  material  to  improve  the  specific  combining  ability  of  a 
known  inbred. 

SOME  GENETIC  CONCEPTS  OF  HETEROSIS 
It  seems  very  evident  to  the  writer  that  heterosis,  the  increased  vigor  of  F] 
over  the  mean  of  the  parents  or  over  the  better  parent,  whichever  definition 
is  "used,  is  not  due  to  any  single  genetic  cause.  A  brief  summary  of  various 

TABLE  3.14 

DISTRIBUTION  OF  %  MOISTURE  AND  YIELD  OF  32  So 
PLANTS  OF  A25XG.  KING  CROSSED  TO  A73XA375. 
CLASSES  OF  L.S.D.  5%  AROUND  MEAN  OF  A25X 
TESTER 


%ear 
mois. 


+2 
+  1 

1 

3 

2 

-1 

-2 

2 
3 

8 

5 

5 
3 

—  2 

-1 

+  1     +2 

(mean  of  A25X 
tester) 


Yield 
(mean  of  A25X  tester) 


TABLE  3.15 

PERFORMANCE  INDICES  OF  So  AND  Si  LINES 
FROM  A25XG.  KING  WHEN  CROSSED  TO 
A73XA375  TESTER  AND  COMPARED  WITH 
A25XTESTER 


So 

Si 

Gamete 

No.  OF 

Number 

Si's 

1947 

1949 

1949 

19  H 

+  11 

+  19 

+25 

5 

20  H 

+  14 

+  9 

+  14 

7 

36  H 

+  9 

+  16 

+  11 

7 

5L 

-11 

-  3 

+  5 

7 

29  L 

-11 

-   1 

-  0 

1 

46  L 

-  5 

+  1 

+  2 

7 

62  H.  K.  HAYES 

theories  advanced  to  explain  heterosis  seems  desirable  to  set  the  stage  for 
later  discussions.  Bruce  (1910)  explained  heterosis  on  the  combined  action 
of  favorable  dominant  or  partially  dominant  factors,  based  as  Richey  (1945a) 
has  emphasized  on  mathematical  expectations. 

Keeble  and  Pellew  (1910)  used  a  similar  hypothesis  on  a  di-hybrid  basis 
to  explain  hybrid  vigor  in  peas.  East  and  G.  H.  Shull  (1910-1914)  believed 
vigor  was  dependent  on  heterozygosis  on  the  basis  that  the  stimulus  of  hy- 
bridity  was  not  entirely  Mendelian.  A.  F.  Shull  (1912)  preferred  the  explana- 
tion that  heterosis  was  due  to  a  stimulus  resulting  from  a  changed  nucleus 
on  a  relatively  unaltered  cytoplasm.  Jones  (1917)  restated  Bruce's  concept 
and  added  the  concept  of  linkage. 

Collins  (1921)  and  Richey  (1945)  have  pointed  out  that  where  large  num- 
bers of  factor  pairs  are  involved  it  would  be  very  difficult  to  recover  all  fac- 
tors in  a  favorable  condition  in  F2,  or  in  later  segregating  generations.  With 
multiple  factors  involved,  however,  linkage  must  of  necessity  make  the  re- 
combination of  factors  more  difficult.  East  (1936)  presented  a  Mendelian 
concept  of  the  interaction  of  alleles  at  the  same  locus  to  explain  heterosis, 
where  two  alleles  of  a  particular  gene  pair  had  each  developed  a  divergent 
physiological  function.  The  writer  believes  he  continued  also  to  accept  the 
previous  explanation  that  heterosis  was  dependent  on  the  cumulative  effect 
of  dominant  or  partially  dominant  linked  genes. 

Gustafsson  (1947),  Hull  (1945a),  Jones  (1945),  Castle  (1946),  and  others 
have  emphasized  the  importance  of  interallelic  action  in  relation  to  heterosis. 
Castle  has  suggested  also  that  the  effect  of  interallelic  action  of  a  single  pair 
of  genes  "is  similar  to  that  of  the  killer  mutation  of  Sonneborn,  except  that 
the  action  induced  in  the  dominant  gene  by  its  sensitized  recessive,  instead 
of  being  harmful,  in  this  case  is  beneficial." 

In  certain  cases  a  homozygous  recessive  pair  of  genes  may  completely 
modify  the  normal  expression  of  either  a  homozygous  or  heterozygous  or- 
ganism. Homozygous  dwarfs  in  maize  condition  such  a  result.  A  cross  be- 
tween two  different  dwarfs,  however,  releases  the  inhibition  of  each  dwarf 
and  results  in  marked  heterosis.  Both  dominant  factors,  where  two  dwarfs 
are  crossed,  appear  to  be  necessary  to  condition  normal  development.  In  this 
case  the  dominant  conditions  of  both  factor  pairs  act  as  complementary  fac- 
tors for  normal  growth. 

It  is  evident  that  genes  are  greatly  affected  in  their  expression  by  differ- 
ences in  both  external  and  internal  environment.  Cytoplasmic  inheritance  of 
male  sterility  may  be  used  for  illustrative  purposes.  Several  cases  of  male 
sterility  in  sugar  beets  and  onions,  for  example,  are  known  that  are  due  to 
maternal  cytoplasmic  inheritance  which  may  be  modified  in  expression  by 
the  dominant  or  recessive  condition  of  one  or  more  factor  pairs. 

Recently  Hsu  (1950)  at  Minnesota  has  studied  the  effect  of  two  pairs  of 
dwarf  factors  of  maize  in  their  homozygous  dominant  and  recessive  condi- 


DEVELOPMENT  OF  THE   HETEROSIS  CONCEPT 


63 


tions,  and  also  when  heterozygous  in  near  isogenic,  homozygous,  and  highly 
heterozygous  backgrounds. 

The  factor  pair  for  Didi  was  studied  in  the  near  isogenic  background  of 
inbred  A 188,  that  of  D^dx  in  the  near  isogenic  background  of  A95-344,  and 
both  factor  pairs  were  studied  in  crosses  between  A 188  X  A95.  Particular 
attention  was  given  to  total  dry  matter  produced  at  various  periods  of  growth 
under  field  conditions  and  to  the  growth  in  length  of  the  coleoptile  and  meso- 
cotyl under  controlled  laboratory  conditions. 

One  comparison  of  the  growth  of  the  mesocotyl  during  a  12-day  period 
for  DiDi  and  Didi  on  three  different  near  isogenic  backgrounds  will  be  con- 
sidered: the  near  isogenic  background,  A188,  and  the  highly  heterozygous 
backgrounds  of  A 188  X  A95  in  the  presence  of  D^D^  and  D^d^,  respec- 
tively. While  Di  conditioned  greater  growth  of  mesocotyl  in  length  than  d\, 
Dx  conditioned  less  development  of  the  mesocotyl  in  length  than  dx. 

The  mesocotyl  length  of  six  strains  consisting  of  comparisons  of  DiDi 
with  Didi  on  three  dififerent  backgrounds  was  taken  as  100.  The  comparisons 
are  summarized  in  Table  3.16  and  in  Figure  3.1. 

It  is  apparent  that  the  superiority  of  DiDi  over  Didi  in  mesocotyl  length 
becomes  less  in  the  highly  heterozygous  background  than  in  the  homozygous 
background  of  A188.  This  may  be  more  evident  from  the  diagram  in  Fig- 
ure 3.1. 

TABLE  3.16 

COMPARATIVE  LENGTH  OF  MESOCOT- 
\T.  FOR  SIX  STRAINS  OF  CORN 


Background 

Percentage 
Difference  in 

Mesocotyl 

Length,  DiDi 

minus  Didi 

Percentage 
Expression  of 
Background 

A188 

19 

16 

4 

89 

A188XA95Z?xZ?x... 
^.\2,%X^9SDxdx.... 

101 

no 

It  seems  of  some  interest  that  the  differences  between  DiDi  and  Didi  were 
smaller  in  the  highly  heterozygous  background  than  in  the  homozygous 
background,  and  that  in  the  presence  of  D^x  that  the  differences  were 
further  reduced  over  those  in  the  presence  of  DxDx.  It  may  be  well  to  recall 
that  dx  conditioned  greater  length  of  mesocotyl  than  Dx- 

Reference  may  be  made  to  an  explanation  by  Torssell  (1948)  of  tlie  decline 
in  green  weight  or  length  of  stem  in  alfalfa  in  different  generations  of  in- 
breeding. It  was  not  greatest  in  the  first  inbred  generation.  He  suggests  there 
was  a  surplus  of  vigor  genes  in  a  heterozygous  condition  in  the  early  genera- 
tions of  selfing,  and  that  great  loss  of  vigor  was  not  observed  until  about  I3 


C/) 


<  -o 


(/) 


(A 


105 


(/) 

Q 

(/) 

Ixl 

O 

tr 

1- 

Q. 

X 

o 

z 

UJ 

h- 

o 

< 

< 

_J 

1- 

UJ 

z 

rr 

UJ 

o 

or 

UJ 

CL 

95 


85 


75 


A    D,D, 


/ 


/ 


/ 


/ 


/ 


/ 


A' 


A 


B 

i± 


C 


80  90  100  110 

PERCENTAGE     EXPRESSION     OF 
BACKGROUND 


Fig.  3.1 — Relative  expression  of  A  A  vs.  A^i  regarding  final  length  of  mesocotyl  on  vari- 
ous backgrounds:  (A,  A188;  B,  A188  X  A95-344  carrying  AA;  C,  A188  X  A95-344  D,d^). 


DEVELOPMENT  OF  THE  HETEROSIS  CONCEPT  65 

when  selfing  reduced  the  necessary  genes  below  a  stage  needed  by  the  or- 
ganism. The  following  quotation  from  Thorssell  emphasizes  the  viewpoint 
that  the  relative  importance  of  genes  controlling  heterosis  is  greatly  in- 
fluenced by  other  factors  of  the  organism: 

The  cumulative  effect  of  heritable  characters,  however,  brings  it  about  that  develop- 
ment, that  is  to  say  green  weight,  does  not  stand  in  arithmetical  proportion  to  the  number 
of  pairs  of  the  dominant  genes  in  question.  From  this  it  follows  also  that  the  said  number  can 
be  reduced  within  a  certain  limit  without  perceptible  or  any  great  influence  upon  green 
weight.  If  this  limit  is  exceeded,  a  considerable  degeneration  sets  in. 

The  speaker  has  chosen  to  consider  heterosis  as  the  normal  expression  of 
a  complex  character  when  the  genes  concerned  are  in  a  highly  heterozygous 
condition.  x'Vs  most  normal  characters  are  the  end  result  of  the  action,  reac- 
tion, and  interaction  of  countless  numbers  of  genes,  and  as  gene  mutation 
constantly  occurs  although  relatively  infrequently,  it  may  be  impossible  to 
obtain  all  essential  genes  in  the  most  favorable  homozygous  state.  After 
selecting  the  best  homozygous  combinations,  further  vigor  will  be  obtained 
due  to  heterozygous  combinations  of  factors.  Dominance  or  partial  domi- 
nance seems  of  great  importance  as  an  explanation  of  hybrid  vigor.  In  some 
cases  there  may  be  extra  vigor  correlated  with  the  heterozygous  condition  of 
pairs  of  alleles.  The  types  of  response  of  inter  and  intra  allelic  factor  interac- 
tions are  without  doubt  dependent  upon  both  external  and  internal  environ- 
ment. 


M.  M.  RHOADES 

University  of  Illinois 


Chapter  4 

Preferential 
Segregation  in  Maize 


The  outstanding  example  of  the  utilization  of  heterosis  in  plant  improve- 
ment is  that  of  hybrid  corn.  Extensive  studies  on  maize  genetics  have  clearly 
demonstrated  that  chromosome  and  gene  segregation  are  in  accordance  with 
Mendel's  laws  of  segregation  and  recombination.  It  would  appear,  therefore, 
that  any  unusual  mechanism  operating  in  maize  to  produce  deviations  from 
normal  Mendelian  behavior  should  be  worthy  of  our  consideration,  even 
though  the  principles  involved  have  no  bearing  on  the  nature  or  manifesta- 
tion of  heterosis.  The  purpose  of  this  section  is  to  present  data  on  preferential 
segregation  in  maize  and  to  offer  a  tentative  interpretation  of  this  phe- 
nomenon. 

Two  kinds  of  chromosome  10,  the  shortest  member  of  the  haploid  set  of 
ten,  are  found  in  populations  of  maize.  The  common  or  normal  type  gives 
typical  Mendelian  ratios  when  the  two  homologues  are  heterozygous  for 
mutant  loci.  The  second  kind  of  chromosome  10,  which  has  been  found  in  a 
number  of  races  from  Latin  America  and  the  southwestern  United  States, 
also  gives  normal  Mendelian  ratios  for  chromosome  10  loci  in  plants  homozy- 
gous for  this  chromosome.  This  second  or  abnormal  kind  of  chromosome  10 
differs  from  the  normal  chromosome  10  by  a  large,  chiefly  heterochromatic 
segment  of  chromatin  attached  to  the  end  of  the  long  arm  and  also  in  the 
chromomeric  structure  of  the  distal  one-sixth  of  the  long  arm  (see  Fig.  4.1 
and  Fig.  1  of  Plate  I).  As  is  illustrated  in  Figure  4.1  the  chromomeres  in  this 
region  are  larger  and  more  deeply  staining  than  are  the  correspondingly 
situated  chromomeres  of  the  normal  homologue. 

Although  normal  Mendelian  ratios  are  obtained  for  segregating  loci  in 
chromosome  10  in  both  kinds  of  homozygotes,  we  were  able  to  show  in  an 
earlier  paper  (Rhoades,  1941)  that  preferential  segregation  occurs  at  mega- 

66 


n 


Fig.  4.1 — Camera  lucida  sketch  at  pachynema  of  bivalent  consisting  of  one  normal  and  one 
abnormal  chromosome  10.  Note  the  dissimilarity  in  chromomere  pattern  in  the  distal  one- 
sixth  of  the  long  arm.  The  identical  chromomere  pattern  found  in  the  remainder  of  the 

chromosomes  is  not  figured  here. 

Fig.  4.2 — Anaphase  I  of  cell  illustrated  in  Figure  4  of  Plate  I.  Some  of  the  disjoining  dyads 
are  normal  appearing  while  others  have  active  neo-centric  regions. 

Fig.  4.3 — Metaphase  I  with  eleven  dyads.  Five  of  the  dyads  have  precocious  neo-centro- 
meres  at  sub-terminal  portions  of  their  long  arms. 

Fig.  4.4 — Anaphase  II  of  cell  illustrated  in  Figure  7  of  Plate  II.  In  some  of  the  inverted 
V-shaped  monads  the  true  centric  regions  are  attracted  toward  the  opposite  pole. 


68  M.  M.  RHOADES 

sporogenesis  in  plants  heterozygous  for  a  normal  and  an  abnormal  type  of 
chromosome  10.  Approximately  70  per  cent  of  the  functioning  mega  spores 
possessed  the  abnormal  10  instead  of  the  usual  50  per  cent.  The  excess  of 
female  gametes  with  the  abnormal  10  was  not  due  to  lethal  factors  or  to 
megaspore  competition.  The  disjunction  of  the  two  dyads  comprising  the 
heteromorphic  bivalent  at  anaphase  I,  and  of  the  two  monads  of  each  dyad 
at  anaphase  II,  was  such  that  an  abnormal  10  chromosome  tended  to  pass 
with  a  high  frequency  to  the  basal  spore  of  the  linear  set  of  four. 

The  factor  or  factors  responsible  for  this  preferential  segregation  reside 
in  the  chromatin  segments  which  differentiate  the  two  kinds  of  chromosome 
10.  Whether  the  distal  one-sixth  of  the  long  arm  or  the  large  heterochromatic 
piece  of  extra  chromatin  carries  the  causative  genes  for  preferential  segrega- 
tion has  not  yet  been  determined — since  these  two  regions  of  the  abnormal 
chromosome  10  have  never  been  separated  by  crossing  over.  The  locus  of 
the  gene  R  is  in  the  long  arm  of  chromosome  10.  There  is  approximately  1  per 
cent  recombination  between  R  and  the  end  of  the  long  arm  in  plants  hetero- 
zygous for  the  two  kinds  of  chromosome  10;  but  every  crossover  distal  to  R 
occurred  to  the  left  of  the  dissimilar  chromomeres  in  the  distal  one-sixth  of 
the  long  arm.  Apparently  little  or  no  crossing  over  takes  place  here,  although 
pairing  at  pachytene  is  intimate. 

Strictly  terminal  chiasmata  in  the  long  arm  have  not  been  observed  at 
diakinesis  in  heterozygous  plants.  The  close  linkage  of  the  R  locus  with  the 
extra  segment  of  abnormal  10  is  due  to  a  suppression  of  crossing  over  in  the 
end  regions  of  the  long  arm.  E.  G.  Anderson  (unpublished)  has  studied  a  re- 
ciprocal translocation  involving  normal  10  with  the  break  distal  to  R,  and 
found  5  per  cent  recombination  between  R  and  the  translocation  point. 
There  is  an  undetermined  amount  of  crossing  over  between  the  translocation 
point  and  the  end  of  the  chromosome.  It  should  be  possible  to  locate  the  re- 
gion or  regions  in  abnormal  10  responsible  for  preferential  segregation  by  ob- 
taining successively  larger  terminal  deficiencies,  but  this  has  not  been  at- 
tempted. 

The  dissimilarity  in  chromomere  pattern  in  the  distal  portion  of  the  long 
arms  of  the  abnormal  and  normal  chromosomes  10,  together  with  the  lack  of 
crossing  over  in  this  region,  suggest  the  possibility  that  the  gene  content  may 
not  be  identical  in  the  two  kinds  of  chromosome  10.  Inasmuch  as  plants 
homozygous  for  the  abnormal  chromosome  10  are  not  noticeably  different  in 
growth  habit  and  general  appearance  from  sibs  carrying  only  the  normal  10, 
it  would  appear  that  some  kind  of  structural  modification  was  responsible  for 
the  suppression  of  crossing  over.  To  assume  that  this  distal  region  consists 
of  non-homologous  loci  in  the  two  types  of  chromosome  would  mean  that 
plants  with  two  abnormal  10  chromosomes  would  be  homozygous  deficient 
for  certain  loci  found  in  the  comparable  region  of  normal  10.  This  appears 
unlikely. 


PREFERENTIAL  SEGREGATION   IN  MAIZE  69 

That  a  structural  difference,  aside  from  the  extra  chromatin  of  abnormal 
10,  exists  between  the  two  kinds  of  chromosome  10  also  is  indicated  by  the 
pairing  relationships  in  plants  trisomic  for  chromosome  10.  in  plants  with 
two  normal  and  one  abnormal  chromosome  10,  trivalent  associations  were 
observed  in  251  (60.2  per  cent)  among  a  total  of  417  microsporocytes.  When 
a  chain  of  3  was  found  at  diakinesis,  the  abnormal  10  occupied  a  terminal 
position  in  90  per  cent  of  the  cells.  It  was  united  with  a  normal  chromosome  10 
by  a  chiasma  in  the  short  arm.  A  univalent  chromosome  10  was  found  at 
diakinesis  in  39.8  per  cent  of  the  pollen  mother  cells. 

If  pairing,  as  reflected  by  chiasmata  formation,  were  random  among  the 
three  chromosomes,  the  ratio  of  normal -.abnormal  chromosomes  10  in  the 
univalent  class  should  be  2 : 1.  Actually  the  unpaired  chromosome  was  a  nor- 
mal 10  in  28  cells  among  a  total  of  166,  while  in  the  remaining  138  cells  the 
univalent  was  an  abnormal  10.  In  individuals  again  trisomic  for  chromo- 
some 10,  but  possessing  one  normal  and  two  abnormal  chromosomes,  the 
percentage  of  trivalent  associations  at  diakinesis  was  57.9  in  a  total  of  513 
cells.  In  the  chains  of  3,  the  two  abnormal  homologues  were  adjacent  mem- 
bers, joined  by  a  chiasma  between  their  long  arms,  in  70  per  cent  of  the 
cases.  An  unpaired  chromosome  10  was  found  in  42.1  per  cent  of  the  micro- 
sporocytes. 

If  pairing  were  random,  two  times  as  many  abnormal  lO's  as  normal  lO's 
should  be  found  as  univalents;  but  in  a  total  of  216  cells  an  abnormal  10 
was  the  univalent  in  69,  while  a  normal  chromosome  10  was  the  univalent 
in  147.  Chiasma  formation  among  the  three  chromosomes  10  of  trisomic 
plants  clearly  is  not  at  random.  There  is  a  marked  preference  for  exchanges 
in  the  long  arm  between  the  two  structurally  identical  homologues.  If  synap- 
sis usually  begins  at  the  ends  and  progresses  proximally,  the  non-random  as- 
sociations found  in  trisomic  plants  become  understandable.  Normal  recom- 
bination values  for  the  li-gi  and  gi-R  regions  which  lie  proximal  to  R  (see 
Table  4.1  for  gi-R  data)  indicate  that  any  suppression  of  crossing  over  is 
confined  to  the  region  beyond  the  R  locus  in  disomic  plants  heterozygous  for 
the  two  kinds  of  chromosome  10.  It  is  no  doubt  significant  that  differences 
in  chromomeric  structure  are  not  found  in  regions  proximal  to  the  R  locus. 

Inasmuch  as  the  R  locus  is  closely  linked  with  the  extra  chromatin  of  ab- 
normal 10,  the  ratio  oi  R:r  gametes  from  heterozygous  plants  gives  a  good 
approximation  of  the  frequency  with  which  the  abnormal  chromosome  passes 
to  the  basal  megaspore.  The  genetic  length  of  the  long  arm  of  chromosome  10 
is  such  that  at  least  one  chiasma  is  found  in  the  arm.  If  one  chiasma  invari- 
ably occurs  in  the  long  arm  of  heteromorphic  bivalents,  each  of  the  two  dis- 
joining dyads  of  anaphase  I  will  possess  one  normal  chromatid  and  one  ab- 
normal chromatid.  Preferential  segregation  would  be  restricted  to  the  sec- 
ond meiotic  division,  and  occur  only  if  the  orientation  of  the  dyad  on  the 
spindle  of  metaphase  II  were  such  that  the  abnormal  chromatid  passed  to 


70  M.  M.  RHOADES 

the  lower  pole  of  the  spindle.  Normal  segregation  would  occur  in  those  mega- 
sporocytes  which  had  homomorphic  dyads. 

If  the  terminal  segment  of  abnormal  10  determines  preferential  segrega- 
tion, it  follows  that  loci  near  the  end  of  the  long  arm  will  be  preferentially 
segregated  more  frequently  than  loci  further  removed  from  the  end  of  the 
chromosome.  From  the  data  in  Tables  4.1  and  4.2  it  is  evident  that  the  dis- 
tortion from  a  1 : 1  ratio  is  greater  for  the  R  locus  than  for  the  more  proximal- 
ly  situated  gi  locus.  The  li  locus  which  is  proximal  to  gi  was  less  affected 
than  ^1. 

Longley  (1945)  reported  non-random  segregation  at  megasporogenesis  for 
chromosome  pairs  other  than  chromosome  10  when  one  of  the  two  homologues 
had  a  prominent  knob  and  the  other  was  knobless.  Segregation  was  random 
for  these  heteromorphic  bivalents  in  plants  homozygous  for  the  normal  chro- 
mosome 10,  and  non-random  if  abnormal  10  was  heterozygous.  He  studied 
preferential  segregation  of  chromosomes  9  and  6.  The  data  for  chromosome  9 
are  the  most  instructive.  Some  strains  of  maize  have  a  chromosome  9  with  a 
knob  at  the  end  of  the  short  arm,  others  have  a  knobless  chromosome  9.  The 
C,  Sh,  and  Wx  loci  lie  in  the  short  arm  of  this  chromosome,  with  Wx  nearer 
to  the  centromere.  C  and  Sh  are  in  the  distal  one-third  of  the  short  arm.  Ap- 
proximately 44  per  cent  recombination  occurs  between  Wx  and  the  terminal 
knob — they  approach  independence — while  C  and  Sh  are  23  and  26  recombi- 
nation units  distant  from  the  knob.  * 

When  plants  of  knob-C/knobless-c  constitution,  which  were  also  heterozy- 
gous for  abnormal  10,  were  pollinated  by  recessive  c,  64  per  cent  of  the  func- 
tioning megaspores  possessed  the  C  allele.  The  Sh  locus,  close  to  C,  showed  a 
similar  degree  of  preferential  segregation  in  comparable  tests,  but  the  Wx 
locus  was  little  affected.  Such  a  progressive  decrease  in  effect  is  expected  if 
the  terminal  knob  on  the  short  arm  is  instrumental  in  producing  preferential 
segregation.  The  part  played  by  the  knob  of  chromosome  9  was  wholly  un- 
expected. Obviously  this  heterochromatic  structure  can  no  longer  be  con- 
sidered as  genetically  inert.  The  data  on  various  loci  in  chromosomes  9  and 
10  prove  that  the  degree  of  preferential  segregation  of  a  locus  is  a  function 
of  its  linkage  with  heterochromatic  regions  which,  in  some  way,  are  con- 
cerned with  non-random  segregation. 

The  data  presented  above  show  that  alternative  alleles  are  not  present  in 
equal  numbers  among  the  female  gametes  when  abnormal  10  is  heterozygous. 
We  have  here  an  exception  to  Mendel's  first  law.  Are  deviations  from  Men- 
del's second  law,  the  independent  assortment  of  factor  pairs  on  non-homolo- 
gous chromosomes,  also  occurring?  This  question  is  answered  by  Longley's 
data  where  the  C  and  R  loci  are  both  segregating  preferentially.  In  separate 
experiments  he  found  the  C  locus  was  included  in  64  per  cent  and  the  R  locus 
in  69  per  cent  of  the  functioning  megaspores.  Assuming  that  these  percent- 
ages hold  in  plants  where  both  are  simultaneously  segregating,  the  observed 


•fppl 


2 


*   ^< 


^  i' 


•»  ' .' 


4 


Plate  I:  Fig.  1 — Pachytene  showing  homozygous  abnormal  10.  Carmine  smear.  The  proximal  jjortion  of 
the  extra  chromatin  is  euchromatic  as  is  a  smaller  distal  piece.  A  large  and  conspicuous  knob  lies  between 
the  two  euchromatic  portions.  Fig.  2 — Metaphase  I  in  microsporocyte  homozygous  for  abnormal  10. 
Carmine  smear.  The  ten  bivalents  each  have  their  true  centric  regions  co-oriented  on  the  spindle.  The 
onset  of  neo-centric  activity  is  manifest  in  the  second,  sixth,  and  seventh  bivalents  from  the  right.  The 
third  and  fourth  bivalents  from  the  right  are  somewhat  superimposed.  Figs.  3  and  4 — .\naphase  I  in  mi- 
crosporocyte homozygous  for  abnormal  10.  Carmine  smear.  Some  of  the  dyads  are  undergoing  a  normal 
anaphase  separation  while  in  others  the  neo-centric  regions  are  pulling  the  ends  poleward.  Note  that 
the  normal  ap[)earing  dyads  are  slower  in  their  poleward  migration.  F'^ig.  4.2  is  a  drawing  of  Fig.  4  above. 


Plate  II:  Figs.  1  and  2 — Metaphase  II  in  plant  homozygous  for  abnormal  10.  Carmine 
smear.  Precocious  poleward  movement  of  neo-centric  regions  is  clearly  evident.  One  dyad 
has  a  single  neo-centric  region  (Fig.  4.5,  dyad  No.  8)  while  the  left-most  dyad  has  a  neo- 
centric  region  in  both  long  arms  (Fig.  4.5,  dyad  No.  7).  This  cell  was  figured  in  Rhoades 
and  Vilkomerson  1942.  Figs.  3  and  4 — .Anaphase  II  in  jilant  homoz\gous  for  abnormal  10. 
Carmine  smear.  Note  that  the  rod-shaped  monads  with  precocious  neo-centromeres  are 
the  first  to  reach  the  poles.  Fig.  5 — Metaphase  II  in  plant  homozygous  for  abnormal  10. 
Carmine  smear.  The  only  chromosome  of  the  haploid  complement  which  can  be  recognized 
at  metaphase  II  is  chromosome  6  which  has  a  satellite  at  the  end  of  the  short  arm.  In  this 
cell  the  chromosome  6  d\'ad  is  the  second  from  the  left.  That  the  terminal  chromosome  of 
the  satellite  is  actuall\-  a  small  knob  is  indicated  by  the  formation  of  neo-centric  regions  al 
the  end  of  the  short  arm.  Fig.  6 — Early  anaphase  II  in  plant  heterozygous  for  abnormal  10. 
Carmine  smear.  That  the  poleward  movement  of  neo-centric  regions  is  less  rapid  in  hetero- 
zygous than  in  homozygous  abnormal  10  plants  is  indicated  here  by  the  relatively  slight 
attenuation  of  the  rod-shaped  monads.  Fig. — 7  Late  anaphase  II  in  plant  homozygous  for 
abnormal  10.  Carmine  smear.  The  previously  greatlx-  stretched  rod  monads  with  precocious 
neo-centromeres  have  contracted.  Note  the  inverted  V-shaped  chromatids.  This  is  the  same 
cell  shown  in  Figure  4.4.  Fig.  8 — Side  view  of  metaphase  I  in  a  normal  plant  showing  the 
fibrillar  nature  of  the  chromosomal  fibers.  Fi.xed  in  Benda,  stained  with  haemotoxylin. 
Paratiine  section.  The  only  chromosomal  fibers  j^resent  are  those  formed  by  the  true  cen- 
tromeres. Ordinarily  chromosomal  fibers  are  not  evident  in  carmine  smears  since  they  are 
destroyed  by  acetic-alcohol  fi.xation  and  it  is  nccessar\'  to  use  special  techniques  to  demon- 
strate them.  Similar  fibrillar  chromosomal  fibers  are  found  at  neo-centric  regions  when 
proper  fixation  and  staining  methods  are  employed.  Fig.  9  (top) — Polar  view  of  meta- 
phase I  in  normal  plant.  Fixed  in  Benda,  stained  with  haemotoxylin.  Parafhne  section. 
Note  the  arrangement  of  the  ten  bivalents  on  the  ecjuatorial  plate.  This  microsporocyte 
was  cut  slightl}'  above  the  metaphase  plate.  The  next  section,  which  includes  the  remaining 
portion  of  this  cell,  is  a  cross  section  through  the  ten  sets  of  chromosomal  fibers. 


.// 


J 


> 


/ 


j> 


1 


/ 


'         2 

'\   h  I- 
't  J.^f 

i    .  S 


■*l!7 


^.-■IH.-  ;^^'4«H*' 


# 


8 


// 


T^      •    >         .' 


A  V  V, 


^        A. 


'  M»  /• 


••» 


1 


^lMjJ 


TABLE  4.1 

LINKAGE  DATA  FROM  THE  CROSS  OF  G  r  ABNORMMVg  R 
NORM.\L  X  gr  d'd' 


Linkage 
Phase 

Constitution  of  Chromosomes 

Repulsion 

(0) 

G 
r 

(o) 

g 
R 

(X) 

G 
R 

(x) 
S 

r 

Total 

Ratio  of 
i?:r  ON  Ear 

243 
102 
150 
396 
154 
169 
215 
231 

138 
86 

114 
50 
81 
90 
61 
79 

29 
9 

18 
7 

11 
21 
24 
35 

49 
13 
20 
59 
29 
30 
77 
81 

459 
210 
302 
512 

275 
310 
377 
426 

186/?:326r 
136/?:319r 
145/?:288r 
169/?:588r 
120/?:277r 
127/?:223r 
102/?:338r 
133/?:358r 

1660 

699 

154 

358 

2871 

1118/?:2717r 

%Rm  total  =  29.7  %  g  in  total  =  36.8 

%  i?  in  non-crossover  classes  =  29.6 
%  /J  in  crossover  classes  =  30.1 

G  —  R  recombination  =  17.8% 


29.2%  R 


TABLE  4.2 

LINKAGE  DATA  FROM  THE  CROSS  OF  G  r  NORMAL/g  R 

ABNORMAL  X  gr 


Linkage 
Phase 

Constitution  of  Chromosomes 

Repulsion 

(o) 
G 
r 

(o) 

g 
R 

(x) 
G 
R 

(x) 

g 

r 

Total 

Ratio  of 
R\r  (m  Ear 

12 
38 
35 
39 

87 

96 

86 

107 

13 
29 

21 

1 

6 

7 
9 

113 
169 
161 
176 

182/?:  42r 
188/?:   59r 
230/?:   74r 
241/?:   77r 

124 

376 

96 

23 

619 

841/?:252r 

%  r  seeds  in  total  =  23.8 

%  r  seeds  in  non-crossover  classes  =  24.8 

%  T  seeds  in  crossover  classes  =  19.3 

G  —  R  recombination  =  19.2% 


72  M.  M.  RHOADES 

frequencies  of  F2  phenotypes  can  be  compared  with  those  calculated  on  the 
assumption  of  independent  assortment.  The  two  values  agreed  very  closely, 
indicating  little  or  no  deviation  from  the  law  of  independent  assortment. 
His  data,  from  plants  where  loci  in  chromosomes  9  and  6  are  both  segregat- 
ing preferentially,  likewise  permit  such  a  conclusion  to  be  drawn. 

In  my  1942  paper  on  preferential  segregation  the  statement  was  made 
that  the  chromosomes  in  plants  with  the  abnormal  chromosome  10  formed 
extra  chromosomal  (half  spindle)  iibers  at  regions  other  than  the  true  centro- 
mere region.  Rhoades  and  Vilkomerson  (1942)  found  these  supernumerary 
chromosomal  fibers  were  produced  only  in  plants  homozygous  or  heterozy- 
gous for  the  abnormal  10,  and  that  sister  plants  homozygous  for  the  normal 
10  had  chromosomal  fibers  originating  solely  from  the  localized  centric  re- 
gion in  an  orthodox  manner  (see  Fig.  8  of  Plate  II).  Although  the  abnormal 
chromosome  10  was  clearly  responsible  for  the  formation  of  these  neo-centric 
regions,  they  were  not  restricted  to  this  chromosome  since  many  of  the  non- 
homologous chromosomes  had  supernumerary  chromosomal  fibers.  The  ab- 
normal chromosome  10  is  thus  responsible  for  the  formation  of  neo-centric 
regions,  as  well  as  for  preferential  segregation.  Since  1942,  a  considerable 
body  of  data  has  been  obtained  bearing  on  the  behavior  of  abnormal  10. 
Some  of  the  more  pertinent  observations  have  suggested  a  cytological  mecha- 
nism for  the  phenomenon  of  preferential  segregation. 

The  unorthodox  formation  of  supernumerary  chromosomal  fibers  from  neo- 
centric  regions  is  limited  to  the  two  meiotic  divisions.  (For  a  description  of 
normal  meiosis  in  maize  see  Rhoades,  1950.)  The  first  meiotic  division  is  in 
no  way  exceptional  until  metaphase  I  is  reached.  Normal  appearing  bivalents 
are  co-oriented  on  the  spindle  figure  in  a  regular  manner  with  the  half  spindle 
fibers,  arising  from  the  true  centric  regions,  extending  poleward.  Normally 
these  fibers  effect  the  anaphase  movement  of  the  disjoining  dyads  with  the 
localized  centromere  region  leading  the  journey  to  the  spindle  pole.  How- 
ever, in  plants  with  the  abnormal  10,  chromosomal  fibers  arise  from  distal 
regions  of  the  chromosome  while  the  bivalents  are  still  co-oriented  on  the 
spindle  at  metaphase  I.  The  neo-centric  regions  are  drawn  poleward  more 
rapidly  than  the  true  centric  regions.  Consequently  the  distal  ends,  instead 
of  being  directed  toward  the  spindle  plate  during  anaphase  I,  lead  the  way 
to  the  pole. 

The  appearance  of  many  disjoining  dyads  at  anaphase  I  suggests  that 
their  poleward  migration  is  due  largely,  even  exclusively,  to  the  fibers  origi- 
nating from  the  neo-centric  regions.  The  primary  centric  region  appears  to 
play  no  active  role  even  though  it  possessed  chromosomal  fibers  at  meta- 
phase I  when  the  tetrad  (bivalent)  was  co-oriented.  At  mid-anaphase  there  is 
no  indication  of  the  presence  of  these  fibers  in  many  of  the  dyads  with  the 
precocious  neo-centric  regions. 

Figure  4.5  and  Figures  3  and  4  of  Plate  I  illustrate  some  of  the  observed 


4 


ANAPHASE  I  DYADS 


A 

11 


7  8 

METAPHASE  H  DYADS 


J      ii      I      ^ 

12        13  »U     ^15 


10 


i 


12 


ANAPHASE  IE  MONADS 


t 


16 


Fig.  4.5 — All  figures  are  from  carmine  smears  of  homozygous  abnormal  10  plants.  Figures 
1-5  represent  various  configurations  found  at  anaphase  I.  Figure  1  is  a  normal  dyad  with 
chromosomal  fibers  formed  only  at  the  true  centric  region.  In  Figure  2,  two  arms  have 
formed  neo-centric  regions.  The  true  centric  regions  appear  to  be  inactive.  Figure  3  shows  a 
dyad  with  two  neo-centric  regions  and  an  active  true  centric  region  whose  chromosomal 
fibers  are  directed  away  from  the  nearest  pole.  Figure  4  is  a  dyad  with  a  single  neo-centric 
region.  In  Figure  5  the  two  neo-centric  regions  are  directed  to  opposite  poles.  Figures  6-7 
illustrate  various  metaphase  II  dyads.  The  location  of  the  equatorial  plate  is  represented 
by  horizontal  lines.  Figure  6  is  essentially  normal  with  no  formation  of  neo-centromeres. 
Figure  7  is  a  dyad  with  two  neo-centric  regions  directed  toward  opposite  poles.  There  is  a 
single  neo-centric  region  in  Figure  8.  Figure  9  is  a  dyad  which  is  disjjlaced  from  the  equa- 
torial plate.  The  true  centric  region  has  divided  to  form  two  independent  monads.  Each 
monad  has  formed  two  neo-centric  regions  which  are  oriented  toward  opposite  poles.  In 
Figure  10  one  of  the  monads  has  its  two  neo-centromeres  directed  to  opposite  poles.  Fig- 
ures 11-16  are  illustrations  of  monads  found  at  anaphase  II.  Figure  11  is  a  normally  dis- 
joining monad.  In  Figure  12  a  single  neo-centromere  is  evident.  Figure  13  shows  two  neo- 
centric  regions.  Figure  14  has  a  single  neo-centromere  which  was  active  at  metaphase  II. 
In  Figure  15,  chromosomal  fibers  have  arisen  from  two  neo-centric  regions  and  also  from 
the  true  centric  region.  The  true  centric  region  and  the  neo-centromeres  are  acting  in  op- 
posite directions.  Figure  16  shows  a  monad  with  two  neo-centric  regions  which  are  directed 
toward  opposite  poles.  This  type  of  monad  is  derived  from  those  shown  in  Figure  9. 


74  M.  M.  RHOADES 

anaphase  I  configurations.  Chromosomal  fibers  may  arise  from  one  or  both 
of  the  long  arms  of  each  dyad  at  late  metaphase  or  early  anaphase  I.  Al- 
though it  was  not  always  possible  to  differentiate  between  long  and  short 
arms,  the  neo-centric  regions  in  general  appear  to  be  confined  to  the  long 
arm.  When  both  long  arms  of  the  two  chromatids  of  a  dyad  possessed  a  neo- 
centric  region,  the  chromosomal  fibers  arising  from  these  centric  regions  were 
usually  directed  toward  the  same  pole.  Occasionally  they  were  oriented  to 
opposite  poles  thus  causing  a  great  attenuation.  In  such  cases,  however, 
those  chromosomal  fibers  nearest  to  one  pole  were  powerful  enough  to  over- 
come the  oppositely  directed  force  of  the  second  neo-centromere.  Despite  the 
great  complexity  of  configurations  at  anaphase  I  resulting  from  interacting 
and  conflicting  half-spindle  fibers  arising  from  both  the  true  and  neo-centric 
regions,  the  end  of  anaphase  I  usually  finds  ten  dyads  at  each  pole.  Some- 
times, however,  greatly  stretched  chromosomes  undergo  breakage.  This 
breakage  doubtless  accounts  for  the  higher  pollen  abortion  (about  10  per 
cent)  found  in  homozygous  abnormal  10  plants  as  contrasted  to  the  lower 
(0-5  per  cent)  pollen  abortion  of  normal  sibs. 

Even  though  one  or  two  arms  of  some  dyads  are  markedly  stretched  at 
anaphase  I,  the  ensuing  telophase  is  normal.  All  four  arms  of  each  dyad  con- 
tract to  form  a  spherical  mass  of  chromatin  which  is  loosely  enveloped  by 
the  lightly-staining  matrical  substance.  The  chromonemata  uncoil  during 
interphase  and  early  prophase  II  finds  each  daughter  cell  with  ten,  long  X- 
shaped  dyads  of  typical  appearance.  The  two  chromatids  comprising  each 
dyad  are  conjoined  by  the  undivided  primary  centric  region.  There  is  no  indi- 
cation of  neo-centric  regions,  although  some  of  the  long  arms  possessed  chro- 
mosomal fibers  at  the  preceding  anaphase. 

The  onset  of  metaphase  II  sometimes  occurs  before  the  dyads  have  under- 
gone their  usual  contraction.  Occasionally  chromosomal  fibers  arising  from 
neo-centric  regions  in  the  long  arms  are  found  at  late  prophase  II.  These 
precociously  acting  fibers  produce  an  extension  of  the  long  arms  before  any 
spindle  is  visible.  This  observation  is  of  singular  importance.  Some  authori- 
ties believe  that  the  centromere  region  is  attracted  (whatever  this  term  may 
signify)  to  the  spindle  pole.  Here  we  have  a  movement  produced  by  the 
chromosomal  fibers  of  neo-centric  regions  in  the  absence  of  an  organized 
spindle.  The  way  in  which  these  neo-centric  fibers  act  can  only  be  conjec- 
tured, but  no  interaction  between  centric  regions  and  spindle  pole  is  essential. 
It  is,  indeed,  probable  that  the  only  role  of  a  bipolar  spindle  is  to  provide  a 
structural  frame  which  channels  the  chromosomes  to  the  spindle  poles. 
Clark's  (1940)  studies  on  divergent  spindles  are  pertinent  in  this  respect. 

The  objection  may  be  raised  that  the  chromosomal  fibers  of  neo-centric 
regions  are  not  comparable  to  those  arising  from  the  true  centric  region.  I 
do  not  believe  this  is  a  valid  criticism.  Not  only  are  both  kinds  of  fibers  con- 
cerned with  chromosome  movement,  but,  as  will  be  shown  in  a  later  section, 


PREFERENTIAL  SEGREGATION  IN  MAIZE  75 

the  fiber-producing  activity  of  the  neo-centric  regions  is  a  ])r()<luct  of  the  true 
centric  region. 

The  appearance  of  neo-centric  fibers  in  prophase  II  is  not  the  rule. Usually 
the  dyads  come  to  lie  with  the  true  centric  region  on  the  spindle  plate  at 
metaphase  II  before  any  pronounced  activity  of  neo-centric  regions  is  ap- 
parent. Before  the  primary  centric  region  divides,  thus  permitting  a  normal 
anaphase,  chromosomal  fibers  again  arise  near  the  distal  ends  of  the  long 
arms  of  some  dyads.  These  newly  formed  fibers  move  the  long  arms  poleward 
while  the  dyad  is  still  held  on  the  metaphase  plate  by  the  undivided  true 
centric  region.  This  poleward  movement  is  so  rapid  that  the  ends  of  the 
chromosomes  may  reach  the  spindle  poles  before  the  true  anaphase  occurs. 
Eventually  the  true  centric  region  becomes  functionally  split,  and  the  two 
monads  fall  apart  and  pass  poleward.  It  is  evident  from  Figures  4.4  and  4.5 
and  Figure  7  of  Plate  II  that  the  configurations  of  the  disjoining  monads 
(chromatids)  at  anaphase  II  are  greatly  different  from  normal. 

Neo-centric  activity,  as  shown  by  formation  of  additional  chromosomal 
fibers,  occurs  in  plants  both  homozygous  and  heterozygous  for  the  abnormal 
10,  but  it  is  much  more  striking  in  homozygous  plants.  Plants  trisomic  for 
abnormal  10  were  not  greatly  different  from  homozygous  disomic  sibs. 

Precocious  chromosomal  fiber  formation  by  neo-centromeres  at  metaphase 
II  appears  in  general  to  be  confined  to  the  long  arms  of  the  dyads,  although 
it  is  often  difficult  to  differentiate  between  two  unequal  arms  when  one  is 
stretched  poleward.  Some  chromosomes  have  arm  ratios  so  extreme  that 
the  distinction  between  long  and  short  arms  is  clear,  and  in  these  chromo- 
somes the  precocious  fibers  at  metaphase  II  arise  from  the  long  arms.  It  is 
perhaps  significant  that,  with  the  exception  of  the  terminal  knob  on  the  short 
arm  of  chromosome  9,  all  remaining  knobs  in  our  material  were  situated  in 
the  long  arms.  (Chromosome  6  had  two  small  knobs  in  its  long  arm  but  a 
maximum  of  one  knob  was  present  in  the  other  chromosomes.)  Corre- 
spondingly, only  one  of  the  two  arms  of  any  chromatid  had  neo-centric 
activity  at  metaphase  11.^  The  number  of  dyads  with  precocious  spindle 
fibers,  as  judged  by  the  number  of  arms  pulled  poleward  at  metaphase  II, 
varied  in  different  strains.  The  maximum  number  in  some  plants  was  seven, 
in  others  five,  etc.  Plants  with  seven  knobbed  chromosomes  had  a  maximum 
of  seven  dyads  with  arms  stretched  poleward  at  metaphase  II.  Those  with 
four  knobs  had  four  such  dyads.  That  is,  a  strong  correlation  exists  between 
knob  number  and  the  number  of  dyads  with  neo-centric  activity  at  meta- 
phase II. 

A  further  observation  of  some  interest  was  that  in  plants  homozygous  for 
all  knobs  both  homologous  arms  of  a  dyad  usually  were  pulled  poleward  at 
metaphase  II;  while  in  plants  heterozygous  for  some  knobs  many  of  the 
dyads  had  only  one  arm  with  neo-centric  activity  (see  Figure  4.5  and  Figures 

1.  With  the  possible  exception  of  chromosome  6.  See  Figure  5  of  Plate  II. 


76  M.  M.   RHOADES 

1  and  2  of  Plate  II).  It  is  not  unreasonable  to  assume  that  dyads  with  both 
homologous  arms  exhibiting  neo-centromeres  at  metaphase  II  carried  a  knob 
in  each  chromatid,  while  dyads  with  one  neo-centromere  consisted  of  one 
knobbed  and  one  knobless  chromatid.  Such  heteromorphic  dyads  would  arise 
from  heteromorphic  bivalents  by  a  crossover  between  the  true  centromere 
and  the  knob.  We  believe  that  only  knobbed  chromatids  have  active  neo- 
centromeres  at  metaphase  II,  and  that  knobless  ones  are  normal  at  this  stage. 
Unfortunately,  knobs  cannot  be  recognized  at  metaphase  II,  and  the  validity 
of  the  above  assumptions  rests  upon  indirect  but  convincing  evidence. 

Two  types  of  disjoining  monads  are  found  at  anaphase  II,  those  which 
are  rod-shaped  and  those  which  are  V-shaped.  Monads  which  had  one  arm 
extending  poleward  at  metaphase  II  are  rod-shaped.  They  are  the  first  to 
reach  the  pole.  Indeed  distal  portions  of  such  chromatids  already  had  arrived 
there  during  metaphase  II  owing  to  the  early  action  of  their  neo-centromeres. 
The  V-shaped  monads  of  anaphase  II  are  derived  from  those  chromatids 
devoid  of  neo-centromeres  at  metaphase  II.  The  poleward  migration  of  some 
monads  is  first  begun  by  the  chromosomal  fibers  emanating  from  the  true 
centric  region,  but  shortly  after  anaphase  is  initiated  chromosomal  fibers 
may  arise  from  the  ends  of  both  arms.  These  terminally  placed  fibers,  which 
are  directed  to  the  same  pole,  propel  their  ends  poleward  with  such  rapidity 
that  the  ends  first  overtake  and  then  pass  the  centric  region  in  the  course  of 
anaphase  migration.  Consequently  these  monads  reach  the  poles  as  inverted 
V-shaped  chromosomes  (see  Fig.  4.4).  The  spindle  fibers  from  the  true  centric 
region  now  are  directed  toward  the  spindle  plate  rather  than  to  the  pole — they 
have  reversed  their  orientation.  This  would  be  impossible  if  chromosomal 
fibers  were  of  a  thread-like  structure.  It  is  more  likely  that  these  fibers  repre- 
sent nothing  more  than  lines  of  force  emanating  from  the  centromere.  In- 
verted V-shaped  chromatids  are  not  invariably  found  at  anaphase  II. 

Some  monads  have  chromosomal  fibers  only  at  the  true  centric  region  and 
move  poleward  in  a  normal  fashion.  Either  neo-centric  regions  are  not  pres- 
ent, or  else  arise  too  late  to  be  effective.  It  should  be  emphasized  that  a  funda- 
mental distinction  exists  between  the  rod  and  inverted  V  chromatids  found 
at  anaphase  II.  The  rod-shaped  monads  come  from  dyads  with  neo-centric 
activity  at  metaphase  II.  Their  supernumerary  chromosomal  fibers  arise 
from  one  arm.  Their  sub-terminal  location  suggests  they  may  arise  adjacent 
to  the  knob,  but  this  is  merely  a  conjecture.  The  later-formed  extra  chromo- 
somal fibers  of  the  inverted  V  chromatids,  which  are  knobless,  are  terminal 
and  arise  from  both  arms. 

If  a  dyad  is  oriented  on  the  spindle  plate  at  metaphase  II  before  the  onset 
of  precocious  neo-centromere  activity,  the  supernumerary  chromosomal 
fibers  arising  from  the  knobbed  arm  of  the  chromatid  situated  slightly  above 
the  spindle  plate  are  directed  toward  the  upper  (nearest)  pole,  and  those 
from  the  bottom  chromatid  go  to  the  lower  pole — they  are  co-oriented  (see 


PREFERENTIAL  SEGREGATION   IN  MAIZE  11 

Fig.  4.3).  No  such  regularity  is  found  in  those  infrequently  occurring  dyads 
which  are  longitudinally  displaced  from  the  spindle  {)late  at  metaphase  II. 
Their  true  centric  regions  divide  prematurely.  Consequently,  the  two 
chromatids  of  these  displaced  dyads  no  longer  remain  conjoined,  but  fall 
apart  to  become  independent  monads  which  lie  side-by-side,  parallel  with 
the  longitudinal  axis  of  the  spindle. 

The  neo-centric  activity  which  these  monads  now  manifest  is  similar  to 
that  found  at  anaphase  II  for  those  monads  derived  from  normally  oriented 
dyads  lacking  precocious  neo-centromeres  at  metaphase  II,  in  that  neo- 
centromeres  may  arise  from  the  ends  of  both  arms.  When  this  occurs,  the 
orientation  of  the  two  neo-centromeres  of  each  monad  is  usually  to  opposite 
poles,  but  sometimes  both  ends  of  a  monad  are  directed  toward  the  same 
pole.  Although  the  monads  from  displaced  dyads  have  neo-centromeres  at 
the  end  of  each  arm,  one  end  being  attracted  to  the  nearest  pole  and  the  other 
to  the  more  distant  pole,  normal  disjunction  usually  occurs.  This  requires 
one  monad  to  move  away  from  the  nearest  pole  toward  which  one  of  its  ends 
is  attracted,  and  to  pass  to  the  more  distant  pole,  while  the  other  monad  goes 
to  the  nearest  pole.  It  is  difficult  to  interpret  this  phenomenon  in  terms  of 
strength  of  attraction  as  a  function  of  distance  from  centromere  to  pole. 

The  formation  of  neo-centric  regions  requires  the  presence  of  the  abnormal 
chromosome  10.  In  its  absence,  no  such  regions  are  found.  It  appears  highly 
probable  that  heterochromatic  knobs  located  on  other  chromosomes  also  are 
concerned  in  the  formation  of  precocious  centric  regions  at  both  meiotic 
metaphases,  since  the  cytological  observations  show  a  correlation  between 
number  of  knobs  and  number  of  precocious  centric  regions.  Knobless  arms 
later  form  neo-centric  regions,  but  not  until  anaphase  movement  has  already 
been  initiated  by  the  true  centric  region. 

It  is  possible  that  maize  chromosomes  possess  latent  centric  regions  which 
are  activated  by  the  abnormal  10.  It  has  been  demonstrated,  however,  that 
the  true  centric  region  is  involved  in  the  formation  of  neo-centromeres. 
Plants  homozygous  for  abnormal  10  and  heterozygous  for  the  long  para- 
centric inversion  in  chromosome  4,  studied  by  McClintock  (1938)  and  Mor- 
gan (1950),  were  obtained.  Both  the  normal  and  inverted  chromosome  4 
carried  a  large  knob  in  the  long  arm  which  is  included  in  the  inverted  seg- 
ment. Single  crossovers  within  the  inversion  give  rise  to  two  non-crossover 
monocentric  chromatids,  one  dicentric  chromatid  which  forms  a  bridge  at 
anaphase  I,  and  an  acentric  fragment.  The  knobbed  acentric  fragment  lies 
passively  on  the  spindle  with  no  indication  of  spindle  fiber  activity.  Neo-cen- 
tromeres arise  from  the  same  chromatin  segments  comprising  the  acentric 
fragment  when  they  constitute  a  portion  of  a  whole  chromosome  4.  It  fol- 
lows that  the  true  or  primary  centromere  plays  an  essential  role  in  the  pro- 
duction of  neo-centromeres. 

The  localized  centromeres  of  maize  chromosomes  are  concerned  with  the 


78  M.  M.  RHOADES 

elaboration  of  fiber-producing  material.  Normally  this  unique  substance  is 
confined  to  the  true  centric  region,  hence  chromosomal  fibers  arise  solely 
from  this  part  of  the  chromosome. 

It  is  our  belief:  (1)  that  these  centric  regions  produce  an  over-abundance 
of  fiber-forming  material  if  abnormal  10  is  present  in  the  nucleus;  (2)  that  a 
portion  of  this  substance  escapes  from  the  confines  of  the  centric  regions  and 
moves  distally  along  the  chromosome  to  produce  supernumerary  chromo- 
somal fibers;  and  (3)  that  the  knobs  either  stimulate  centric  activity  or  else 
cause  the  excess  fiber-forming  substance  to  move  preferentially  along  knob- 
bearing  arms  so  that  neo-centric  activity  is  first  manifested  by  these  arms. 

The  failure  of  the  acentric  fragment  to  form  chromosomal  fibers  suggests 
that  the  postulated  movement  of  the  material  from  the  true  centric  region 
occurs  after  crossing  over  has  taken  place.  If  it  happened  prior  to  pachytene, 
the  regions  which  later  constitute  the  acentric  fragments  would  receive  some 
of  this  fiber-producing  substance  which  subsequently  could  form  spindle 
fibers.  In  support  of  the  above  interpretation  is  the  observation  that  small 
aggregations  of  a  substance  similar  in  appearance  to  that  located  in  the  true 
centric  region  are  sometimes  found  near  the  distal  regions  of  some  chromo- 
somes at  metaphase  I  and  metaphase  II.  This  observation  is  subject  to  vari- 
ous interpretations.  But  in  conjunction  with  the  behavior  of  acentric  frag- 
ments, it  strengthens  the  hypothesis  that  the  production  of  neo-centromeres 
is  intimately  related  to  the  presence  or  activity  of  the  primary  centric  region. 
It  is  obvious  that  the  presumed  movement  of  the  products  of  the  centromere 
along  the  arms  of  the  chromosome  has  a  bearing  on  the  kinetic  theory  of  Posi- 
tion Effect. 

Evidence  has  been  presented  that  the  abnormal  chromosome  10  produces 
the  phenomenon  of  preferential  segregation,  and  that  it  also  causes  the  for- 
mation of  neo-centromeres.  Are  these  two  phenomena  related — does  prefer- 
ential segregation  occur  as  a  consequence  of  neo-centric  activity?  While  no 
definite  answer  can  be  given  at  this  time  a  tentative  hypothesis  has  been  de- 
veloped. Sturtevant  and  Beadle  (1936),  seeking  to  account  for  the  absence  of 
egg  and  larvae  mortality  following  single  crossovers  in  paracentric  inversions 
in  Drosophila,  postulated  that  the  crossover  chromatids  were  selectively 
eliminated  from  the  egg  nucleus.  The  two  spindles  of  the  second  meiotic  divi- 
sion in  Drosophila  eggs  are  arranged  in  tandem.  Following  a  crossover  within 
the  inverted  segment,  the  tetrad  at  metaphase  I  consists  of  two  non-crossover 
chromatids,  a  dicentric  and  an  acentric  chromatid. 

They  assumed  that  the  chromatin  bridge  arising  from  the  dicentric  chro- 
matid, when  the  homologous  centromeres  pass  to  opposite  poles  at  anaphase 
I,  ties  its  two  centromeres  together.  The  spatial  arrangement  thus  produced 
is  such  that  the  two  monocentric  chromatids  lie  nearer  the  two  poles  than 
does  the  dicentric  chromatid. 

The  persistence  of  this  relationship  into  the  second  division  results  in  a 


PREFERENTIAL  SEGREGATION  IN  MAIZE  79 

non-random  orientation  on  the  metaphase  II  spindles.  The  monocentric,  non- 
crossover  chromatids  are  free  to  pass  to  the  two  terminal  poles,  while  the  two 
centromeres  from  the  dicentric  chromatid  are  directed  to  the  two  inner  poles. 
Consequently,  at  anaphase  II  the  terminal  poles  each  receive  a  non-crossover 
chromatid.  Since  the  egg  nucleus  arises  from  the  innermost  terminal  pole  it 
would  contain  a  non-crossover  chromatid  with  a  full  set  of  genes.  The  cor- 
rectness of  this  ingenious  hypothesis  was  established  by  Darlington  and  La 
Cour  (1941)  in  Lilium  and  Tulipa  and  by  Carson  (1946)  in  Sciara. 

It  is  possible  that  a  somewhat  similar  mechanism  is  operating  in  Zea  to 
produce  preferential  segregation.  In  maize,  as  in  Drosophila,  the  two 
spindles  of  the  second  meiotic  division  of  megasporogenesis  are  arranged  in 
a  linear  order.  The  basal  megaspore  of  the  linear  set  of  four  develops  into 
the  female  gametophyte,  the  remaining  three  aborting.  We  know  that  in 
plants  heterozygous  for  knobbed  and  knobless  chromosomes,  one  arm  of 
some  of  the  disjoining  dyads  at  anaphase  I  possess  precociously-acting 
chromosomal  fibers  not  present  in  the  homologous  arm.  There  is  reason  to 
believe  that  the  knobbed  arms  form  precocious  neo-centromeres  while  knob- 
less  arms  do  not.  Owing  to  the  rapidity  with  which  neo-centric  regions  pass 
poleward  at  anaphase  I,  those  chromatids  with  neo-centromeres  reach  the 
pole  in  advance  of  knobless  arms  lacking  neo-centromeres.  In  a  dyad  con- 
sisting of  one  knobbed  and  one  knobless  chromatid,  the  knobbed  chromatid 
would  come  to  lie  closer  to  the  pole,  while  the  knobless  one  would  face  the 
spindle  plate. 

In  order  to  account  for  preferential  segregation,  it  is  necessary  to  assume 
that  this  orientation  persists  until  the  second  metaphase,  and  that  it  results 
in  the  knobbed  chromatids  facing  the  two  terminal  poles  while  the  two  knob- 
less ones  would  be  oriented  toward  the  two  inner  poles.  On  such  a  mechanism, 
preferential  segregation  would  occur  only  when  a  crossover  takes  place  be- 
tween the  knob  and  the  true  centromere  in  a  heterozygous  bivalent.  The 
extent  of  preferential  segregation  would  be  a  direct  function  of  the  amount  of 
crossing  over  in  the  knob-centromere  region. 

Such  an  explanation  can  only  be  considered  as  a  working  hypothesis.  It 
can  be  critically  tested,  however,  and  such  experiments  are  being  conducted 
by  Jean  Werner  Morgan,  who  also  participated  in  the  studies  reported  here. 
They  include  varying  the  crossover  distance  between  knob  and  centromere 
by  translocation  and  inversion,  testing  for  preferential  segregation  of  hetero- 
morphic  chromosomes  other  than  chromosome  10  in  plants  homozygous  for 
abnormal  10,  determining  neo-centric  activity  in  chromatids  with  knobs  in 
both  the  long  and  short  arm,  etc.  I  prefer  not  to  mention  her  incomplete 
findings  at  this  time,  since  to  do  so  would  detract  from  continued  interest  in 
her  work. 

The  phenomenon  of  preferential  segregation  is  by  no  means  confined  to 
maize.  Sturtevant  (1936)  found  a  non-random  segregation  of  three  chromo- 


80  M.  M.  RHOADES 

somes  IV  in  Drosophila.  Bridges,  in  Morgan,  Bridges,  and  Sturtevant  (1925), 
established  that  the  distribution  of  the  chromosomes  in  triploid  Drosophila 
was  not  according  to  chance.  Beadle  (1935)  reported  that  crossing  over  in 
triploid  Drosophila  near  the  centromere  region  between  one  member  of  at- 
tached -X's  and  a  free  X  chromosome  was  correlated  with  autosomal  dis- 
junction. Lower  crossover  values  were  found  in  1X2,4  and  XX  L4  combina- 
tions than  in  IX  1.1  and  XX  2.4  gametes.  This  non-random  distribution 
indicates  a  correlated  orientation  of  non-homologous  chromosomes  on  the 
equatorial  plate. 

In  Sciara  the  paternal  set  of  chromosomes  moves  away  from  the  pole  of 
the  monocentric  spindle  of  the  primary  spermatocyte.  The  two  sister  .Y 
chromosomes  pass  to  the  same  pole  at  the  second  spermatocyte  division 
(Metz,  1938).  Schrader  (1931)  observed  a  non-random  orientation  in  Pro- 
tortonia  which  led  to  selective  distribution  in  secondary  spermatocytes. 
Catcheside  (1944),  in  an  analysis  of  Zickler's  data  on  spore  arrangement  in 
the  Ascomycete  Bombardia  lunata,  found  that  certain  genes  were  prefer- 
entially segregated.  Not  all  of  the  above  examples  are  strictly  comparable  to 
the  situations  found  in  maize,  Sciara,  and  Bombardia.  In  the  latter  cases  a 
specific  spindle  pole  receives  a  certain  chromosome  or  set  of  chromosomes, 
while  in  the  Drosophila  cases  particular  chromosomes  pass  preferentially  to- 
gether, but  presumably  at  random,  to  either  pole. 

The  neo-centromeres  arising  from  chromosome  ends,  reported  in  rye  by 
Prakken  and  Muntzing  (1942)  and  Ostergren  and  Prakken  (1946),  closely 
resemble  those  found  in  maize.  In  both  maize  and  rye  the  neo-centric 
regions  are  found  on  arms  possessing  knobs  (heterochromatin),  and  the  pole- 
ward movement  of  neo-centromeres  is  precocious  in  both  plants.  Unfortu- 
nately, nothing  is  known  about  preferential  segregation  in  rye,  but  it  should 
occur  if  our  hypothesis  is  correct. 


R.  A.  BRINK 

University  of  Wisconsin 


Chapter  5 

Inbreeding  and  Crossbreeding 
in  Seed  Development 


It  is  now  generally  recognized  that  the  effects  on  growth  of  inbreeding  and 
crossbreeding  are  intimately  interwoven  in  the  whole  complex  fabric  of 
development  and  reproduction.  Not  only  are  the  effects  widespread  and 
often  of  major  consequence  in  the  economy  of  the  organism,  but  sometimes 
they  are  manifested  in  devious  ways.  Such  is  the  case  in  the  seed  of  flowering 
plants. 

The  success  or  failure  of  seed  development  turns  primarily,  not  on  the 
embryo  which  embodies  the  line  of  descent,  but  upon  an  accessory  organ  of 
reproduction,  the  endosperm.  The  novel  origin  and  sensitivity  of  this  latter 
tissue  to  changes  in  genetic  composition  render  early  seed  development  one 
of  the  critical  stages  in  the  life  cycle  of  flowering  plants.  My  colleague,  D.  C. 
Cooper,  and  I  have  been  exploring  these  relations  during  the  past  decade.  An 
attempt  will  be  made  here  to  review  some  of  the  evidence  upon  which  our 
point  of  view  rests,  and  to  call  attention  to  some  of  the  broader  implications 
of  the  main  facts. 

As  a  means  of  bringing  the  important  aspects  of  the  problem  in  flowering 
plants  into  focus,  seed  development  in  the  angiosperms  and  gymnosperms 
will  be  compared.  Essential  features  of  the  general  hypothesis  by  which  we 
have  been  guided  will  then  be  set  fo"rth.  The  central  role  of  the  endosperm  in 
formation  of  the  angiosperm  seed  and  the  responsiveness  of  this  tissue  to 
variations  in  genetic  composition  will  be  illustrated  by  a  consideration  of  the 
immediate  effects  of  self-  and  cross-fertilization  in  alfalfa.  It  will  then  be 
shown  that  the  means  by  which  the  embryo  in  the  common  dandelion,  an 
autonomous  apomict,  is  nourished  is  of  a  type  which  would  be  expected 
according  to  the  hypothesis  proposed. 

*  Paper  from  the  Deparlment  of  Genetics,  College  of  Agriculture,  University  of  Wiscon- 
sin, No.  432. 

81 


82  R.  A.  BRINK 

An  illustration  will  next  be  given  of  endosperm  failure  as  an  isolating 
mechanism.  Finally,  the  significance  of  the  present  results  for  the  problem 
of  artificially  rearing  embryos  whose  development  in  the  seed  is  blocked  by 
endosperm  disfunction  will  be  pointed  out. 

Complete  literature  citations  are  not  given.  These  may  be  found  in  the 
summary  paper  (Brink  and  Cooper,  1947)  in  which  much  additional  evidence 
bearing  on  the  present  thesis  also  is  presented. 

The  endosperm  is  a  special  structure  intercalated  between  the  female 
parent  and  the  embryo,  serving  to  mediate  the  relations  between  the  two. 
The  tissue  originates  from  the  central  cell  of  the  female  gametophyte,  follow- 
ing a  fertilization  distinct  from  that  giving  rise  to  the  embryo.  The  secondary 
fertilization  is  unusual  in  that  two  identical  haploid  nuclei  of  maternal  origin 
are  united  with  one  contributed  by  the  pollen.  The  endosperm  thus  becomes 
3x  in  chromosome  number  in  contrast  with  the  2x  condition  of  the  embryo 
and  the  mother  plant,  respectively.  Endosperm  and  embryo  carry  the  same 
kinds  of  genes,  but  the  genie  balance  may  be  unlike  in  the  two  tissues  by 
virtue  of  the  double  contribution  to  the  endosperm  from  the  maternal 
parent.  A  further  element  of  genetic  heterogeneity  in  the  seed  arises  from  the 
fact  that  nucellus  and  integuments,  which  are  maternal  structures,  may 
differ  in  genotype  from  the  endosperm  and  embryo  which  they  enclose, 
since  they  belong  to  the  previous  generation. 

These  facts,  of  course,  have  long  been  known.  Certain  of  their  implica- 
tions, however,  are  only  now  becoming  apparent.  Particularly  is  this  true  of 
the  secondary  fertilization  on  which  our  attention  will  be  focussed. 

A  word  should  be  said  at  this  point  concerning  the  manner  in  which  the 
endosperm  should  be  visualized.  Many  are  familiar  with  the  tissue  only  in 
the  mature  seeds  of  species  in  which  the  endosperm  persists  as  a  storage 
organ.  This  condition,  well  known  in  the  cereals,  for  example,  is  exceptional 
among  flowering  plants,  and  represents  a  secondary  adaptation  of  signifi- 
cance mainly  for  the  future  seedling.  In  most  species  the  endosperm  either 
does  not  persist  in  the  fully  developed  seed  or  occurs  therein  as  a  residue 
only.  On  the  other  hand,  the  endosperm  is  regularly  a  prominent  organ  in 
the  juvenile  seed.  It  is  especially  active  directly  following  fertilization,  during 
what  may  be  termed  the  lag  phase  of  embryo  growth.  This  period  is  seldom 
longer  than  a  few  days,  and  varies  according  to  the  species.  In  spite  of  its 
typically  ephemeral  character,  the  endosperm  plays  a  critical  role  in  (1) 
transforming  the  mature  ovule  into  a  young  seed  and  (2)  nourishing  the 
embryo  during  its  initial  period  of  growth.  We  are  here  concerned  with  the 
endosperm  in  these  two  relationships  only. 

THE  SEED  IN  GYMNOSPERMS  AND  ANGIOSPERMS 

It  is  helpful  in  understanding  the  significance  of  the  secondary  fertilization 
to  compare  the  circumstances  of  seed  development  in  the  angiosperms  with 


INBREEDING  AND  CROSSBREEDING  IN  SEED  DEVELOPMENT  83 

those  in  the  other  great  class  of  seed  forming  plants,  the  gymnosi)erms.  A 
secondary  fertilization  does  not  occur  in  the  gymnosperms.  The  endosi)crm 
is  a  haploid  tissue  derived  from  the  megas])ore  by  continuous  cell  division. 
The  tissue  is  a  part  of  the  gametophyte  rather  than  an  integral  structure 
distinct  from  both  gametophyte  and  sporophyte,  as  in  the  angiosperms. 

On  the  other  hand,  the  endosperms  in  the  two  classes  of  seed  plants  have 
an  important  common  function,  namely,  nourishment  of  their  respective 
associated  embryos.  The  genetic  equipment  with  which  the  two  kinds  of 
endosperms  are  furnished  differs  in  a  fundamental  respect.  That  of  the 
gymnosperm  is  a  sample  half  of  the  mother  plant's  inheritance,  whereas  the 
angiosperm  endosperm,  being  of  biparental  derivation,  has  two  chances  in- 
stead of  only  one  of  receiving  a  physiologically  effective  genie  complement. 
Insofar  as  the  two  tissues  are  autonomous  in  their  functional  properties,  the 
angiosperm  endosperm,  therefore,  is  equipped  to  meet  much  more  exacting 
requirements  than  its  counterpart  in  the  gymnosperms.  A  summary  review 
of  the  differences  in  the  gymnosperm  and  angiosperm  ovules  and  seeds  at 
fertilization,  and  during  the  immediately  subsequent  period,  shows  the  im- 
portance of  (or  necessity  for)  a  secondary  fertilization  in  the  flowering 
plants  in  order  to  maintain  continuity  of  the  life  cycle  at  this  stage. 

The  differences  between  the  mature  ovules  of  gymnosperms  and  angio- 
sperms which  appear  to  have  a  direct  bearing  on  the  present  problem  may  be 
summarized  as  follows: 

1.  The  seed  coat  in  the  gymnosperms  approaches  its  mature  size  at  the 
fertilization  stage.  The  angiosperm  seed  coat  undergoes  extensive  growth  sub- 
sequent to  fertilization.  These  facts  are  of  interest  in  relation  to  the  total 
food  requirements  of  the  two  respective  classes  of  growing  seeds  and  the 
post-fertilization  distribution  of  nutrients  between  the  seed  coat  and  the 
enclosed  tissues. 

2.  The  female  gametophyte  in  the  gymnosperms  is  an  extensively  de- 
veloped multicellular  (multinucleate,  in  some  higher  forms)  structure.  Its 
counterpart  in  the  angiosperms  typically  consists  of  only  seven  cells.  The 
potential  disadvantage  of  the  extreme  reduction  of  the  female  gametophyte 
in  the  flowering  plants  will  be  considered  below. 

3.  Generally  speaking,  the  gymnosperm  ovule  is  rich  in  food  reserves, 
whereas  the  angiosperm  ovule  is  sparsely  supplied.  This  means  that  in  the 
latter,  the  large  volume  of  nutrients  required  for  growth  of  the  endosperm, 
embryo,  and  seed  coat  must  be  moved  in  from  other  parts  of  the  plant.  In 
the  gymnosperms  an  extensive  supply  is  directly  at  hand. 

4.  So  far  as  may  be  inferred  from  the  published  accounts,  fertilization  in 
the  gymnosperms  initiates  a  new  cycle  of  growth  in  the  embryo  only.  Other 
parts  of  the  ovule  do  not  appear  to  be  stimulated.  Double  fertilization  in  the 
angiosperms,  in  contrast,  not  only  marks  the  inception  of  endosperm  and 
embryo  formation,  but  also  incites  pronounced  mitotic  activity  and  en- 


84  R.  A.  BRINK 

largement  of  the  cells  in  the  integuments.  Thus,  with  the  exception  of  the 
nucellus  which  is  broken  down  and  absorbed  by  the  rapidly  expanding  endo- 
sperm, all  the  elements  of  the  young  seed  which  were  previously  quiescent, 
suddenly  spring  into  active  growth  following  syngamy. 

Consideration  of  these  differences  between  the  seeds  of  gymnosperms  and 
angiosperms  led  us  some  ten  years  ago  to  explore  the  hypothesis  that  the 
secondary  fertilization  in  angiosperms  is  essentially  a  means  of  enhancing 
the  competitive  power  of  the  endosperm  relative  to  the  maternal  portions  of 
the  seed — by  conferring  upon  the  endosperm  the  advantages  of  hybridity. 
The  nutritive  requirements  of  the  young  seed  suddenly  are  raised  from  a  low 
to  a  high  level  since  fertilization  starts  a  new  cycle  of  growth  in  the  massive 
integuments.  The  nutrient  supply,  on  the  other  hand,  quickly  falls  to  the 
plane  which  can  be  maintained  by  movement  of  foods  into  the  seed  from 
other  parts  of  the  plant  as  a  result  of  exhaustion  of  the  limited  ovule  reserves. 

It  seemed  reasonable  to  assume  that,  within  the  seed,  the  incoming  nutri- 
ents would  tend  to  be  partitioned  between  the  different  tissues  according  to 
the  respective  amounts  of  growth  occurring  in  them.  On  this  basis,  the  ex- 
tensively developed  integuments  would  consume  the  major  portion.  The 
diminutive  endosperm  and  embryo  would  receive  but  a  small  fraction  of  the 
total.  Under  these  conditions,  failure  of  the  young  seed  through  starvation  of 
the  embryo  could  arise,  unless  the  endosperm — as  the  nutritive  agent  of  the 
embryo — were  endowed  with  special  properties  which  offset  its  initially  small 
size.  It  seemed  essential  that  the  endosperm,  by  one  means  or  another,  be 
enabled  to  quickly  acquire  a  position  of  physiological  dominance  in  the 
juvenile  seed  in  order  to  insure  continued  development. 

Two  genetic  characteristics  of  the  endosperm  suggest  themselves  as  being 
important  in  this  connection.  The  first  is  the  triploid  condition  of  the  nuclei. 
Little  is  known  of  the  physiological  effects  of  ploidy  in  general,  and  virtually 
nothing  of  its  meaning  in  special  situations  of  this  kind.  One  suspects,  how- 
ever, that  the  endosperm  gains  some  advantage  from  its  extra  chromosome 
garniture,  as  such,  in  mediating  the  relations  between  the  diploid  maternal 
parent  and  the  young  diploid  embryo.  It  is  also  probably  significant  that, 
whereas  the  embryo  inherits  equally  from  the  two  parents,  two-thirds  of  the 
endosperm's  genie  complement  is  derived  from  the  plant  upon  which  it  is 
nutritionally  dependent  and  one-third  of  the  complement  from  the  male 
parent. 

Heterozygosis  is  the  second  characteristic  of  the  endosperm  which  might 
enhance  the  inherent  physiological  efficiency  of  this  tissue.  The  possibility  of 
heterozygosity  arises,  of  course,  from  the  biparental  origin  of  the  endosperm 
mother  nucleus.  The  condition  is  realized  in  matings  between  genetically 
different  plants.  Haploidy  of  the  endosperm,  as  occurs  in  the  gymnosperms, 
appears  to  be  genetically  insufficient  for  seed  development  in  the  flowering 
plants.  Early  post-fertilization  circumstances,  particularly  the  dependence 


INBREEDING  AND  CROSSBREEDING  IN   SEED   DEVELOPMENT  85 

upon  and  competition  for  an  outside  nutrient  supply  in  the  latter,  require 
that  the  tissue  shall  share  in  the  advantages  of  sexuality.  The  advantage 
gained  is  not  that  of  amphimixis  in  general,  as  in  the  embryo,  but  solely  the 
extra  vigor  of  growth  associated  with  the  union  of  unlike  nuclei  in  the  mother 
cell.  Thus  hybrid  vigor  in  the  endosperm  has  some  claim  to  uniqueness.  The 
sole  object  gained  by  entry  of  a  sperm  into  the  nuclear  makeup  of  this  sterile 
tissue  is  the  added  vigor  of  growth  thus  acquired.  Some  of  the  evidence  by 
which  the  validity  of  this  point  of  view  may  be  tested  will  now  be  considered. 

INBREEDING  AND  CROSSBREEDING   EFFECT  ON 
SEED  COLLAPSE   IN  MEDICAGO   SATIVA 

Two  classes  of  matings  on  seven  alfalfa  plants  were  carried  out  under 
favorable  growtli  conditions  in  a  greenhouse.  After  removal  of  the  anthers 
from  the  flowers  used,  a  part  of  the  flowers  were  pollinated  with  pollen  from 
the  same  respective  plants.  This  constitutes  the  self-fertilized  series.  Other 
flowers  on  the  same  plants  were  cross-pollinated,  the  pollen  being  derived  in 
each  case  from  an  unrelated  plant  within  the  group.  These  matings  comprise 
the  cross-fertilized  series. 

Since  alfalfa  is  regularly  cross-fertilized,  the  second  series  of  matings  is 
designed  to  maintain  the  level  of  heterozygosity  normal  to  the  endosperm 
and  embryo  in  this  species.  The  enforced  self-fertilization,  on  the  other 
hand,  would  be  expected  to  reduce  heterozygosity  in  the  endosperm  mother 
nucleus  and  the  zygote  by  50  per  cent.  It  is  proposed  to  review  the  conse- 
quences for  seed  development  of  this  sharp  reduction  in  heterozygosis. 

Following  the  above  two  series  of  matings,  the  pistils  were  collected  at  30, 
48,  72, 96, 120,  and  144  hours  and  imbedded  in  paraffin.  After  sectioning  and 
staining,  data  were  taken  on  fertility  of  the  ovules,  frequency  of  fertile  ovules 
collapsing,  number  of  cells  in  the  embryo,  and  number  of  nuclei  in  the  endo- 
sperm. Detailed  observations  were  made  subsequently  on  growth  of  the 
integuments. 

Alfalfa  was  known  previously  to  be  partially  self-incompatible.  It  was 
not  unexpected,  therefore,  to  find  that  only  15  per  cent  of  the  ovules  became 
fertile  after  selfing  in  contrast  to  66  per  cent  after  cross-pollination.  The  new 
fact  which  emerged  was  the  much  higher  incidence  of  collapse  of  ovules  sub- 
sequent to  fertilization  in  the  selfed  than  in  the  crossed  group.  The  data  are 
summarized  in  Table  5.1.  Fertilization  occurred  within  about  30  hours  after 
pollination  under  the  prevailing  conditions.  It  was  somewhat  delayed  after 
selfing.  Little  evidence  of  breakdown  of  the  seeds  was  found  at  48  hours.  In 
the  72  hour  and  subsequent  collections,  however,  the  phenomenon  was  com- 
mon. The  results  presented  in  the  table  cover  the  period  from  72  hours  to 
144  hours,  inclusive,  and  are  based  upon  433  seeds  and  1682  seeds  in  the 
selfed  and  crossed  series,  respectively. 

Growth  of  the  young  seed  at  this  stage  appears  to  be  quite  independent 


86 


R.  A.  BRINK 


of  that  of  its  neighbors  in  the  same  ovary.  Furthermore,  the  quickly  succeed- 
ing secondary  effects  of  fertilization,  such  as  enlargement  of  the  surrounding 
fruit,  are  at  a  minimum.  Studies  on  the  reproductive  physiology  of  the  flower- 
ing plants  are  rendered  difficult  by  the  multiplicity  of  changes  which  are 
eventually  set  in  motion  in  the  tissues  of  the  seed,  the  fruit,  and  the  maternal 
plant  following  fertilization.  The  sequence  and  interrelations  of  the  events 
immediately  subsequent  to  syngamy  are  simpler  to  analyze  than  those  which 
occur  later,  in  view  of  the  fact  that  each  very  young  seed  may  be  considered 
to  behave  independently  of  the  others. 

The  data  in  Table  5.1  show  that,  for  each  of  the  seven  plants  tested,  the 

TABLE  5.1 

FREQUENCY  OF  FERTILE  OVULES  COLL.APSING  IN  SEVEN  .ALFALFA 
PLANTS  FOLLOWING  SELF-  AND  CROSS-FERTILIZATION.  DATA  BASED 
ON  COLLECTIONS  AT  72,  96,  120,  AND  144  HOURS  .\FTER  POLLINATION 
(AFTER  COOPER  AND  BRINK,  1940) 


Self-fertiliz.'\tion 

Cross-fertiliz.ation 

Plant 

No.  of  Fertile  Ovules 

Percentage 
Collapsing 

Plants 
Crossed 

No.  of  Fertile  Ovules 

Percentage 

Selfed 

Total 

Collapsing 

Total 

Collapsing 

Collapsing 

A 

B 

C 

D 

E 

F 

G 

37 
37 
20 
17 
39 
109 
55 

9 

19 

7 

7 

8 

39 

19 

24.3 
51.4 
35.0 
41.2 
20.5 
35.8 
34.5 

AXB 

BXC 

CXD 

DXE 

EXA 

FXG 

GXF 

Total.  . 

187 
110 
171 
171 
146 
228 
198 

13 
5 

13 
16 
9 
14 
16 

7.0 
4.5 
7.6 
9.4 
6.2 
6.1 
8.1 

Total.  . 

314 

108 

34.4 

1211 

86 

7.1 

frequency  of  seeds  collapsing  is  much  higher  in  the  selfed  than  in  the  crossed 
series.  The  proportions  vary  in  different  individuals  from  about  3  to  1  to 
over  11  to  1.  On  the  average,  approximately  five  times  as  many  seeds  con- 
taining inbred  endosperms  and  embryos  collapse  within  the  first  six  days 
after  pollination  as  in  the  crossbred  group.  Since  other  factors  were  not 
varied,  the  decrease  in  survival  in  the  selfed  series  must  be  attributed  to 
the  inbreeding. 

The  evidence,  both  general  and  particular,  points  to  the  endosperm 
rather  than  the  embryo  as  the  seat  of  the  inbreeding  depression  effect.  The 
endosperm  in  alfalfa  is  free  nucleate  up  to  about  144  hours  after  pollination, 
although  it  develops  as  a  cellular  tissue  thereafter.  Successive  waves  of 
mitotic  divisions  traverse  the  tissue,  the  number  of  nuclei  being  doubled  in 
each  cycle.  Thus  growth  during  this  period  proceeds  at  an  exponential  rate. 


INBREEDING  AND  CROSSBREEDING  IN  SEED  DEVELOPMENT  87 

The  concurrent  development  of  the  embryo,  on  the  other  hand,  is  relatively 
slow.  The  zygote  divides  to  form  a  two-celled  proembryo.  Successive  divi- 
sions of  the  apical  cell  give  rise  first  to  a  six-celled  proembr>'o  and  then  to  the 
initials  of  the  definitive  embryo. 

The  pronounced  difference  in  rate  of  development  of  the  two  tissues  is  il- 
lustrated by  the  fact  that  at  144  hours  the  modal  number  of  cells  in  the 
embryo  is  only  16,  whereas  the  typical  number  of  nuclei  in  the  endosperm 
at  this  time  is  128.  Rapid  and  precocious  development  of  the  endosperm  as 
seen  in  alfalfa  is  characteristic  of  the  angiosperms  in  general.  The  much 
higher  level  of  activity  of  the  endosperm  is  presumptive  evidence  tlvit  this 
tissue,  rather  than  the  embryo,  is  especially  subject  to  develo{)mental  upsets 
in  the  young  seed.  Data  available  in  the  present  instance  provide  direct  con- 
firmation of  this  interpretation. 

The  comparative  rates  of  growth  of  endosperm  and  embryo  in  the  selfed 
and  crossed  alfalfa  series  up  to  144  hours  after  pollination  are  illustrated  in 
Figure  5.1.  Not  only  are  the  values  for  the  embryo  low,  but  also  there  is 
little  difference  between  those  for  the  inbred  and  crossbred  series.  The  con- 
clusion appears  warranted  that  the  direct  effect  of  inbreeding  on  the  embryo 
at  this  stage,  if  indeed  there  is  a  demonstrable  effect,  is  too  small  to  account 
for  the  high  frequency  of  seed  collapse.  In  contrast,  there  is  a  very  sharp 
decline  in  rate  of  nuclear  division  in  the  endosperm,  following  enforced  self- 
fertilization  of  this  naturally  cross-fertilized  plant.  The  lower  rate  is  shown 
from  the  first  division  onward.  There  are  about  twice  as  many  nuclei  present 
at  144  hours  in  the  crossbred  as  in  the  inbred  endosperms. 

Due  to  the  partial  self-incompatibility  in  alfalfa,  fertilization  on  the 
average,  is  slightly  delayed  following  selfing.  A  comparison  of  the  rate  of 
growth  of  the  two  classes  of  endosperms  independent  of  time  as  shown  in 
Figure  5.2,  however,  establishes  the  reality  of  the  difference  in  rate  of  growth 
between  the  inbred  and  crossbred  endosperms.  When  the  seeds  are  arrayed 
in  terms  of  cell  numbers  of  the  enclosed  embryo,  it  is  found  that  for  all  nine 
classes  occurring  in  the  material  the  endosperms  are  more  advanced  in  the 
crossbred  than  in  the  inbred  series.  That  is  to  say,  the  embryos  at  a  given 
stage  of  development  have  associated  with  them  more  vigorously  growing 
endosperms  following  cross-fertilization  than  after  selfing.  Moreover,  the 
decrease  in  size  resulting  from  the  inbreeding  is  so  large  that  one  is  led  im- 
mediately to  suspect  that  herein  lies  the  primary  cause  of  the  frequent  seed 
collapse  following  selfing. 

Why  should  impairment  in  rate  of  endosperm  growth  lead  to  arrested  seed 
development?  The  answer  in  the  present  case  is  clear.  As  was  pointed  out 
earlier,  double  fertilization  initiates  not  only  endosperm  and  embryo  develop- 
ment, but  also  a  new  cycle  of  growth  in  the  integuments.  The  latter  compete 
directly  with  the  endosperm  for  the  nutrients  moving  into  the  young  seed. 
If  the  endosperm  is  developing  subnormally,  a  disproportionate  amount  of 


88 


R.  A.  BRINK 


the  incoming  nutrients  is  diverted  to  the  integuments.  As  a  result  this 
tissue  frequently  becomes  hyperplastic.  The  overgrowth  in  the  case  of  al- 
falfa characterizes  the  inner  integument.  As  Dr.  Cooper  observed,  it  begins 
at  a  point  opposite  the  distal  end  of  the  vascular  bundle  where  the  concen- 
tration of  nutrients  maybe  assumed  to  be  the  greatest.  The  inner  integument, 
which  is  normally  two  cell  layers  in  thickness,  becomes  multilayered  and 
somewhat  callus-like  in  the  region  of  the  greatest  mitotic  activity.  This  pro- 
nounced overgrowth  of  the  inner  integument  quickly  reacts  upon  the  endo- 
sperm, further  impairing  its  development.  In  the  seeds  which  fail,  a  complete 


48  72  %  120 

Time  in  hours  after  pollinatior\ 


m 


Fig.  5.1— Increase  in  number  of  cells  in  embryo  and  in  number  of  nuclei  in  endosperm 
following  self-   {broken  line)   and  cross-fertilization   {continuous  line).  After  Brink  and 

Cooper,  1940. 


INBREEDING  AND  CROSSBREEDING  IN   SEED  DEVELOPMENT 


89 


collapse  of  the  endosperm  then  ensues.  Significantly,  breakdown  of  the 
endosperm  tissue  begins  in  the  region  opposite  the  end  of  the  vascular 
bundle  where  the  inner  integument  is  especially  hyperactive.  Following  col- 
lapse of  the  endosperm,  the  young  seed  dies. 

SEED   DEVELOPMENT  WITHOUT  FERTILIZATION 

There  are  a  few  species  of  flowering  plants  in  which  both  endosperm  and 
embryo  develop  without  fertilization.  These  so-called  autonomous  apomicts 


1-celled 
proembri/o 

2-celled 
proem  bryo 

3-celled 
proembryo 

4-celled 
proembryo 

5-ceiied 
proembryo 

6-celied 
proembryo 


2-celied 
emb.*5usp. 


3-ceiled 
emb.-t-5u5p. 

4-celled 
emb.+sasp. 


1   5elf-fcrtilization 
cross-fertilization 


0  10         20         50         40         50         60         70 

Number  of  endosperm  nuclei 

Fig.  5.2 — Number  of  endosperm  nuclei  associated  with  proembryos  and  embryos  at  various 
stages  of  development  following  self-  and  cross-fertilization.  After  Brink  and  Cooper,  1940. 

should  provide  an  independent  test  of  the  hypothesis  that  aggressive  develop- 
ment of  the  endosperm  is  requisite  to  seed  development,  and  that  the  sec- 
ondary fertilization  is  a  device  by  which  aggressiveness  of  the  tissue  is  en- 
hanced. On  the  basis  of  the  reasoning  applied  to  sexual  species,  one  would 
e.xpect  to  find  in  autonomous  apomicts  that  the  embryo  is  not  basically  de- 
pendent on  an  active  endosperm  for  its  nourishment.  So  far  as  I  am  aware, 
the  evidence  bearing  directly  on  this  question  is  limited  to  a  single  study 
which  Cooper  and  I  carried  out  on  the  common  dandelion,  Taraxacum 
officinale  (Cooper  and  Brink,  1949). 

The  common  dandelion  is  triploid  (3x  =  24).  The  regularity  and  abun- 
dance of  seed  production  in  the  plant  is  well  known.  A  full  complement  of  seed 


90  R.  A.  BRINK 

forms  in  the  absence  of  pollination,  as  may  be  demonstrated  easily  by  re- 
moving the  corollas  and  anthers — by  cutting  off  the  distal  portion  of  the 
head  in  the  bud  stage.  Ordinarily  the  anthers  do  not  open  in  the  intact 
mature  flower. 

The  female  gametophyte  is  formed  without  reduction  in  chromosome 
number  of  the  nuclei.  Otherwise  it  is  a  typical  eight-nucleate,  seven-celled 
structure  lying  in  direct  contact  in  the  mature  ovule  with  the  innermost 
layer  of  cells  of  the  single  thick  integument.  The  polar  nuclei  fuse  to  give  a 
hexaploid  primary  endosperm  nucleus.  The  single  layer  of  cells  comprising 
the  nucellus  disintegrates  during  formation  of  the  embryo  sac. 

Sexual  forms  of  the  common  dandelion  are  not  known  to  occur.  Accord- 
ingly another  species,  T.  kok-saghyz,  the  Russian  dandelion,  was  examined 
as  a  control.  T.  kok-saghyz  is  diploid  (2x  =  16)  and,  since  it  is  self-incom- 
patible, requires  cross-pollination  for  seed  formation.  A  comparative  study 
of  T.  officinale  and  T.  kok-saghyz  was  made  with  a  view  to  discovering,  if 
possible,  the  means  by  which  the  former  is  enabled  to  dispense  with  the 
secondary  fertilization,  which  is  essential  to  seed  formation  in  the  latter. 
Heads  were  collected  at  four  stages:  late  bud,  just  prior  to  anthesis,  open 
flower,  and  with  seeds  ranging  up  to  six  days  of  age.  After  sectioning  and 
staining,  the  number  of  cells  in  the  endosperm  and  embryo  was  determined, 
and  observations  were  made  on  the  amount  and  distribution  of  food  ma- 
terials. 

Seed  formation  in  T.  kok-saghyz  follows  the  course  typical  of  the  angio- 
sperms.  Endosperm  and  embryo  development  are  initiated  by  double 
fertilization.  Subsequently,  the  two  tissues  grow  very  rapidly,  and  in  tune 
with  each  other.  Cell  number  in  the  endosperm  increases  exponentially.  The 
endosperm,  however,  is  somewhat  less  precocious  than  in  most  flowering 
plants.  The  seed  is  mature  9-12  days  after  fertilization. 

A  markedly  different  set  of  relations  present  themselves  in  the  seed  of 
the  apomictic  T.  officinale.  The  seed  in  this  species  begins  development  when 
the  flowers  are  in  the  late  bud  stage.  By  the  time  the  flowers  open,  there  may 
be  100  cells  or  more  in  the  endosperm,  the  embryo,  or  in  both  tissues  in  some 
seeds.  A  further  significant  fact  is  the  extraordinary  amount  of  variability 
in  the  size  ratios  of  endosperm  and  embryo  from  seed  to  seed  of  even  age. 
There  is  a  positive  relation  between  cell  number  in  endosperm  and  embryo 
over  the  period  studied — as  would  be  expected  in  view  of  the  fact  that  in 
most  seeds  both  tissues  are  growing.  As  measured  by  the  correlation  co- 
efficient, this  value  is  low  (r  =  .57)  compared  with  that  for  T.  kok-saghyz 
(r  =  .76). 

Average  cell  number  in  the  embryo  in  relation  to  endosperm  size  is  de- 
picted for  the  two  species  in  Figure  5.3.  Cell  number  in  the  endosperm  in- 
creases geometrically,  so  that  size  of  the  tissue  may  be  expressed  appropriate- 
ly in  terms  of  division  cycles.  Embryo  cell  number,  in  contrast,  increases 


INBREEDING  AND  CROSSBREEDING  IN  SEED  DEVELOPMENT 


91 


arithmetically.  It  will  be  noted  from  Figure  5.3  that  the  mean  embryo  cell 
number  in  T.  officinale,  before  the  endosperm  mother  cell  divides  (0  cycle), 
is  about  16.  The  corresponding  value  T.  kok-saghyz  is  1.  This  is  a  reflection  of 
the  fact  that  the  embryo  in  the  apomictic  sj)ecies  usually  starts  growth  in 
advance  of  the  endosperm.  Although  they  start  from  different  levels,  the  two 
curves  are  not  greatly  dissimilar.  The  embryo  in  the  common  dandelion,  on 
the  average,  is  consistently  larger  in  the  young  seed  than  that  of  T.  kok- 
saghyz,  relative  to  given  stages  in  endosperm  develo])ment. 

100 

T.  officinale 

T.   kok-soghyz 


o 

90 

>- 

oc 

m 

60 

z 

UJ 

z 

70 

UJ 

60 

m 

z 

50 

z> 

■^ 

_J 

40 

-I 

UJ 

o 

30 

2 

20 

< 

UJ 

Z 

10 

4 


K.SJ 


0  12  3  4  5  6  7 

NUMBER    OF  DIVISION  CYCLES  IN  ENDOSPERM 

Fig.  5.3 — Early  growth  of  embryo  of  T .  kok-saghyz  and  T.  officinale  in  relation  to  endo- 
sperm size.  After  Cooper  and  Brink,  1949. 

More  instructive  than  the  mean  values  on  which  Figure  5.3  are  based,  is 
the  variability  in  the  frequency  distributions  concerned.  The  data  are  sum- 
marized in  Table  5.2.  A  logarithmic  scale  was  used  in  expressing  embryo 
sizes  merely  as  a  convenient  way  of  summarizing  the  widely  dispersed  values. 
As  mentioned  above,  growth  of  the  embryo  during  this  period  is  approxi- 
mately linear. 

Table  2  reveals  that  the  variability  is  low  in  embryo  cell  number  at  suc- 
cessive stages  of  endosperm  development  in  T.  kok-saghyz.  This  means  that 
embryo  and  endosperm  are  closely  synchronized  in  their  growth  in  the  sexual 
species.  The  variability  in  embryo  size  in  the  apomict,  on  the  other  hand,  is 
enormous.  For  example,  in  seeds  in  which  the  endosperm  is  still  at  the  mother 
cell  stage  (0  cycle),  the  associated  embryos  are  distributed  over  all  size 
classes  from  1  to  128.  The  standard  deviation  for  embryo  cell  number  is 


92 


R.  A.  BRINK 


15.6,  a  value  equal  to  the  mean.  The  range  is  even  greater  in  the  class  of 
seeds  having  128-cell  endosperms,  and  the  standard  deviation  rises  to  51 
cells. 

The  extreme  variability  in  embryo  size  for  given  stages  of  endosperm  de- 
velopment in  T.  officinale  is  a  fact  of  cardinal  importance  in  the  present 
analysis.  Inspection  of  Table  5.2  reveals  certain  details  which  emphasize 
the  significance  of  the  summary  data  on  dispersion.  Note,  for  instance,  that 

TABLE  5.2 
DISTRIBUTION  OF  EMBRYOS  BY  CELL  NUMBER  RELA- 
TIVE TO  ENDOSPERM  DIVISION  CYCLE 
(AFTER  COOPER  AND  BRINK,  1949) 


Endo- 
sperm 

Species 

Total 
Seeds 
Ex- 
amined 

Embryo  Cell  Number — Logarithmic 
Class  Values 

Stand- 
ard 

Division 
Cycle 

1 

2 

4 

8 

16 

32 

64 

128 

256 

Devia- 
tion 

0 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

T.  officinale 
T.  kok-saghyz 

111 
All 

253 

77 

145 

32 

108 

25 

111 
34 

115 
68 

99 

55 

60 
19 

9 

All 

23 
31 

18 

1 

12 

16 

ii 

57 

66 

38 

7 

1 

15.6 
0 

1 

11 
46 

11 
31 

6 

37 

70 

55 

50 

6 

1 

13.6 
0.5 

2 

7 

23 

43 

IZ 

9 

1 

17.0 
0.2 

3 

6 

22 

2 
5 

4 

19 

3 

9 

27 

4 
24 

27 

25 

12 

1 

21.1 
0.7 

4 

4 

1 

39 

2 

23 
40 

7 
10 

1 

40 

14 

2 

19.2 
1.9 

5 

6 

50 
4 

31 
41 

8 
3 

23 

5 

24.2 
4.1 

6 

1 

1 

46 

4 

17 
16 

13 

29.9 

9.0 

7 

2 

1 

28 

3 

51.0 
16.7 

among  the  seeds  still  in  the  endosperm  mother  cell  stage  (0  cycle)  one  con- 
tains an  embryo  in  the  128-cell  class  and  seven  have  embryos  in  the  64-cell 
class.  Similar,  although  less  extreme,  cases  occur  in  the  1 -cycle  and  2-cycle 
endosperm  distributions.  Study  of  the  histological  preparations  shows  that 
the  seeds  in  which  the  embryos  are  found  are  growing  vigorously  and  appear 
capable  of  completing  development.  This  can  mean  only  that  either  very 
small  endosperms  in  T.  officinale  are  extraordinarily  efhcient  structures,  or 
embryo  growth  in  this  species  is  not  dependent  on  an  endosperm. 

At  the  opposite  corner  of  the  table,  on  the  diagonal,  two  seeds  are  entered 
in  the  7-cycle  endosperm  array  in  which  the  embryos  are  still  in  the  one-cell 
stage.  These  seeds  also  appeared  to  be  healthy  and  capable  of  continued 


INBREEDING  AND  CROSSBREEDING  IN  SEED  DEVELOPMENT  93 

development.  These  extreme  examples  {)oint  unmistakably  to  the  conclusion 
that  in  the  apomictic  dandelion  the  endosperm,  as  the  master  tissue  in  the 
young  seed,  has  been  disestablished.  Embryo  growth  must  be  sustained  by 
other  means. 

The  substitute  arrangement  for  nourishing  the  embryo  in  T.  officinale  was 
disclosed  by  a  histological  study  of  the  ovules  of  this  species  and  T.  kok- 
saghyz.  Basically  the  structure  of  the  ovule  is  the  same  in  both.  As  the  female 
gametophyte  expands,  the  nucellus  disintegrates  so  that  the  gametophyte 
comes  to  lie  in  direct  contact  with  the  endothelium  which  comprises  the  in- 
nermost layer  of  cells  of  the  massive  integument.  The  endothelium  [)ersists 
and  appears  to  function  in  the  transfer  of  nutrients  during  the  course  of  seed 
development.  In  T.  kok-saghyz  the  inner  layers  of  integumentary  cells  ad- 
jacent to  the  endothelium  lose  their  contents  during  formation  of  the  game- 
tophyte, and  contain  shrunken  and  misshapen  nuclei  when  the  ovule  is 
mature.  The  cells  of  the  integument  immediately  surrounding  this  depleted 
region  are  densely  cytoplasmic  and  possess  well-defined  nuclei.  The  outer- 
most parenchymatous  cells  of  the  integument  are  highly  vacuolate.  The 
single  vascular  bundle  makes  an  arc  about  the  greatest  circumference  of  the 
ovule  in  both  species.  Only  limited  amounts  of  stainable  reserve  food  ma- 
terials occur  anywhere  in  the  T.  kok-saghyz  ovule. 

The  T.  officinale  ovule  differs  conspicuously  from  that  of  T.  kok-saghyz  in 
possessing  an  abundance  of  reserve  food.  The  cells  of  the  integument  just 
outside  the  endothelium  enlarge  as  the  ovule  matures  and  become  gorged 
with  a  homogeneous  material  which  appears  to  be  proteinaceous  in  composi- 
tion. This  substance  also  extends  between  the  cells  at  the  outer  edge  of  the 
storage  region  proper. 

This  extensive  prestorage  of  protein-rich  food  material  in  the  integument 
provides  an  explanation  of  the  fact  that  embryo  development  in  the  apomict 
may  proceed  normally  in  spite  of  very  limited  endosperm  growth.  The  con- 
ditions render  superfluous  an  aggressively  functioning  endosperm.  The 
embryo  draws  directly  on  a  food  supply  already  at  hand.  From  the  physio- 
logical point  of  view,  the  nutritive  mechanism  in  the  apomict  is  analogous  to 
that  in  the  gymnosperms.  In  both  these  classes  of  plants  certain  of  the 
processes  essential  to  seed  development,  which  follow  double  fertilization  in 
sexual  species  of  flowering  plants,  are  pushed  back  into  the  ovule.  The 
secondary  fertilization,  which  through  its  efifect  on  vigor  of  endosperm  growth 
may  be  looked  upon  as  a  means  of  offsetting  the  tardy  provision  of  nourish- 
ment for  the  embryo,  thus  can  be  dispensed  with. 

SEED   DEVELOPMENT  GRADE  AND  EMBRYO 
GROWTH   POTENTIALITIES 

The  conclusion  that  growth  of  the  angiosperm  seed  is  basically  controlled 
by  the  endosperm  has  an  interesting  corollary.  That  is,  that  the  grade  of  seed 
development  attained  after  a  given  mating  is  not  a  definitive  index  of  the 


94  R.  A.  BRINK 

intrinsic  vigor  of  the  embryo.  This  statement  is  not  intended  to  imply  that 
the  two  phenomena  are  unrelated,  but  rather  that  they  vary  independently 
of  each  other  to  a  significant  degree.  Many  interspecific  matings,  for  example, 
yield  poorly  developed  seeds.  Often  the  embryos  in  these  seeds  give  rise  to 
relatively  weak  plants.  Sometimes,  however,  the  embryos  within  such  seeds 
are  capable  of  forming  plants  of  great  vegetative  vigor.  In  other  words,  the 
fact  that  development  of  the  seed  is  impaired,  even  to  a  degree  that  calls  for 
special  methods  of  propagation,  does  not  necessarily  mean  that  the  embryo  is 
intrinsically  weak.  The  hybrid  during  the  seed  stage  may  merely  be  the 
victim  of  a  faulty  endosperm.  Only  when  released  from  this  stricture  can  the 
inherent  potentialities  of  the  new  individual  be  expressed. 

Two  examples  of  such  intrinsically  vigorous  hybrids  in  which  the  condi- 
tions of  seed  development  have  been  explored  will  be  briefly  mentioned.  They 
differ  in  the  grade  of  seed  development  attained.  Small  but  nevertheless 
germinable  seeds  are  formed  in  the  one  case,  whereas  in  the  other  the  embryo 
egularly  dies  unless  special  precautions  are  taken  to  save  it. 

Cooper  and  I  found  that  when  the  diploid  (2»  =  24)  Red  Currant  tomato, 
Lycopersicon  pimpinelUfolium,  is  pollinated  with  a  particular  strain  of  L. 
peruvianium.,  likewise  a  diploid,  fertilization  occurs  with  high  frequency  but 
all  the  seeds  collapse  before  the  fruit  is  ripe.  Seed  development  follows  a 
familiar  pattern.  The  endosperm  grows  less  vigorously  than  in  normal  L. 
pimpinelUfolium  seeds,  and  the  endothelium  enclosing  it  tends  to  become 
hyperplastic.  Endosperm  cells  become  highly  vacuolate  and  starved  in  ap- 
pearance. Densely  staining  granules  of  unknown  composition  accumulate  in 
the  chalazal  region  just  outside  the  endosperm,  suggesting  that  the  latter 
tissue  is  incapable  of  absorbing  the  available  supply  of  nutrients.  All  the 
seeds  in  the  ripe  fruit  are  shrivelled  and  incapable  of  germination. 

Following  the  application  of  pollen  from  the  same  diploid  strain  of  L. 
peruvianium  to  a  tetraploid  (2w  =  48)  race  of  L.  pimpinelUfolium,  about 
one-half  the  fertile  ovules  develop  into  small  but  germinable  seeds  containing 
triploid  embryos.  The  other  seeds  collapse  at  various  stages  of  growth.  Histo- 
logical examination  of  the  4n  L.  pimpinelUfolium  X  2n  L.  peruvianium  seeds 
shows  retarded  embryo  development  and  a  less  rapid  endosperm  growth 
than  occurs  in  the  normally  pollinated  tetraploid  parent.  The  endosperm  in 
sixteen-day-old  hybrid  seeds  lacks  the  rather  densely  packed  starch  reserves 
characteristic  of  tomato  seeds  at  this  stage.  The  peripheral  layers  of  endo- 
sperm cells  adjacent  to  the  endothelium  break  down.  An  unusually  large 
cavity  is  formed  in  the  interior  of  the  tissue  as  a  result  of  digestion  of  the 
cells  by  the  slowly  differentiating  embryo.  Endosperm  function  is  markedly 
impaired  in  this  cross,  but  in  many  seeds  remains  somewhat  above  the 
threshold  at  which  complete  failure  occurs. 

The  triploid  plants  resulting  from  germinable  4»  L.  pimpinelUfolium  X 
2w  L.  peruvianium  seeds  are  extraordinarily  vigorous.  Although  partially 


INBREEDING  AND  CROSSBREEDING  IN  SEED  DEVELOPMENT  95 

sterile,  they  considerably  exceed  both  the  parents  in  capacity  for  vegetative 
growth.  The  inference  is  clear  that  the  genie  combination  resulting  from  this 
cross  yields  markedly  different  results  in  the  endosperm  and  the  sister 
sporophy  te.  The  difference  in  part  may  be  a  consequence  of  the  2 : 1  balance 
of  L.  pimpinellifolium  and  L.  peruvianium  genes  in  the  embryo  as  compared 
with  the  4: 1  ratio  in  the  endosperm.  The  important  point,  however,  is  that 
the  mechanism  of  seed  formation  in  the  flowering  plants  is  such  that  the  two 
products  of  a  given  double  fertilization  may  be  quite  differently  endowed  in 
terms  of  the  genes  necessary  to  perform  their  respective  functions. 

The  second  example  to  be  discussed  in  this  connection  will  enable  us  to 
visualize  the  limits  which  may  be  reached  in  endosperm  disfunction  with 
retention  of  embryo  viability. 

Fertilization  freely  occurs  when  squirrel-tail  barley,  Hordeum  jubatum  is 
pollinated  by  cultivated  rye,  Secale  cereale.  The  resulting  seeds  all  die,  how- 
ever, within  less  than  two  weeks.  Space  does  not  permit  me  to  recount  here 
the  steps  leading  to  the  breakdown.  They  have  been  described  in  detail  else- 
where (Cooper  &  Brink,  1944;  Brink  &  Cooper,  1944).  The  endosperm  early 
becomes  completely  disorganized.  Some  of  the  embryos  formed,  however, 
reach  a  stage  previous  to  collapse  at  which  time  they  may  be  dissected  from 
the  seed  and  successfully  reared  on  an  artificial  nutrient  medium.  A  single 
plant  was  grown  to  maturity  from  an  embryo  treated  in  this  way.  The 
plant  was  thrifty,  although  sterile.  Representatives  of  the  parent  species 
grown  under  comparable  conditions  were  not  available,  so  that  a  valid  com- 
parison of  relative  vigor  could  not  be  made.  The  hybrid,  however,  appeared 
to  be  intermediate  in  stature  and  number  of  tillers. 

The  extreme  character  of  the  endosperm  disturbances  in  the  H.  jubatum  X 
S.  cereale  seed  indicates  that  this  hybrid  could  not  arise  under  field  condi- 
tions. Although  the  embryo  is  demonstrably  capable  of  continued  develop- 
ment its  growth  is  terminated  in  the  seed  due  to  failure  of  the  associated 
endosperm.  Death  of  the  embryo,  as  an  indirect  result  of  endosperm  disfunc- 
tion following  wide  crosses,  appears  to  be  commoner  than  was  thought  before 
the  physiological  implications  of  the  secondary  fertilization  in  flowering 
plants  were  recognized.  Realization  of  this  fact  has  stimulated  additional 
interest  in  circumventing  the  phenomenon  by  excising  such  embryos  from 
the  seed  and  rearing  them  artificially. 

Artificial  methods  of  cultivating  embryos  removed  from  abortive  seeds 
often  have  been  used  to  extend  the  area  within  which  gene  transfers  may  be 
effected.  Numerous  interspecific  hybrids  have  thus  been  grown  which  other- 
wise are  not  realizable.  The  nature  of  the  general  problem  involved  may  now 
be  seen  in  somewhat  broader  perspective.  Two  points  of  particular  interest 
may  be  noted. 

The  first,  briefly  adverted  to  above,  is  that  the  frequency  with  which  em- 
bryos are  formed  following  matings  between  distantly  related  plants  is  much 


96  R.  A.  BRINK 

higher  than  earlier  believed.  Various  investigators  have  expressed  the  opinion 
that  the  mere  presence  of  growing  pollen  tubes  in  the  style  causes  enlarge- 
ment of  the  ovules.  This  view  now  appears  to  be  incorrect. 

On  the  other  hand,  there  is  a  steadily  increasing  amount  of  evidence  to 
show  that  the  incipient  growth  of  the  ovules,  following  many  interspecific 
matings  which  do  not  yield  functional  seeds,  is  a  response  to  fertilization. 
That  is  to  say,  the  block  in  the  reproductive  cycle  which  was  assumed  to 
intervene  prior  to  fertilization  actually  occurs  following  syngamy.  Embrycs 
are  formed  in  these  cases,  but  they  perish  when  the  young  seed  fails  to  de- 
velop. Some  rather  extreme  examples  of  this  phenomenon  which  have  been 
observed  in  our  laboratory  include  Nicotiana  ghitinosa  X  Petunia  violacea, 
N.  glntinosa  X  Lycopersicon  esculentum,  and  Medicago  saliva  X  M.  scutellata. 

It  is  not  to  be  inferred  that  all  hybrid  embryos  of  this  general  class  are 
capable  of  growing  into  mature  plants.  The  fact  that  the  seeds  containing 
them  collapse  is  not  proof,  however,  of  intrinsic  inviability.  An  unknown  but 
probably  significant  proportion  of  these  novel  zygotic  combinations  are  po- 
tentially propagable.  The  problem  is  to  discover  the  means  by  which  they 
may  be  reared.  This  brings  us  to  the  second  point — the  nature  of  the  problem 
to  be  faced  in  growing  very  small  excised  embryos. 

With  few  exceptions,  the  embryos  which  have  been  successfully  culti- 
vated artificially  have  been  removed  from  the  seed  at  rather  advanced  stages 
of  development.  Unless  they  are  multicellular  and  differentiation  has  at  least 
begun,  the  embryos  usually  do  not  grow  on  the  media  which  thus  far  have 
been  devised.  There  are  reasons  for  thinking  that  the  nutritional  require- 
ments of  these  older  embryos  are  simpler  than  those  in  a  juvenile  condition. 
Histological  evidence  shows  that  at  the  early  stages  of  seed  development  the 
embryo  is  enclosed,  or  nearly  enclosed,  in  the  highly  active,  young  endo- 
sperm. The  endosperm  cells  adjacent  to  the  proembryo  and  the  very  young 
embryo  remain  intact.  A  little  later,  as  the  embryo  enlarges,  these  cells 
begin  to  break  down  and  their  contents  disappear.  Eventually  all  the  endo- 
sperm tissue  is  consumed  in  most  species. 

One  may  infer  from  these  facts  that  the  embryo  is  dependent  upon  the 
endosperm  for  certain  metabolites  which  initially  the  embryo  is  quite  in- 
capable of  synthesizing.  The  endosperm  may  be  pictured  as  secreting  the 
needed  materials  at  the  early  post-fertilization  stage,  and  yielding  them 
later  in  a  more  passive  fashion  as  the  tissue  becomes  lysed.  Meanwhile  the 
embryo  becomes  progressively  less  dependent  upon  the  endosperm  by  acquir- 
ing for  itself  the  synthetic  capabilities  previously  limited  to  the  nurse  tissue. 
On  this  view  the  very  young  embryo  is  an  obligate  parasite  on  the  endo- 
sperm. Once  past  the  state  of  obligate  parasitism,  growth  of  the  embryo  may 
be  effectively  supported  by  comparatively  simple  nutrients  such  as  may  be 
provided  in  artificial  culture  media. 

Visualized  in  those  terms,  the  problem  of  cultivating  very  young,  excised 


INBREEDING  AND  CROSSBREEDING  IN   SEED  DEVELOPMENT  97 

embryos  resolves  itself  into  the  discovery  of  means  of  duplicating  the  un- 
known but  presumably  special  nutritive  functions  of  the  normal  endosperm. 
Two  possibilities  suggest  themselves  in  this  connection.  One  is  to  determine 
natural  sources  of  the  special  metabolites  produced  by  the  endosperm  and 
then  add  these  materials  to  the  nutrient  medium.  Van  Overbeek  (1942)  ob- 
tained significant  improvement  in  the  growth  of  small  Datura  stramonhim 
embryos  by  supplying  them  with  unautoclaved  coconut  milk.  Blakeslee  and 
Satina  (1944)  later  reported  that  the  coconut  milk  could  be  replaced  by  un- 
autoclaved malt  extract.  The  other  possibility  is  to  cultivate  the  embr>^os 
artificially  in  association  with  actively  functioning  endosperm  tissue.  Cur- 
rent findings  offer  some  encouragement  that  the  latter  procedure  may  prove 
efficacious. 

Dr.  Nancy  Ziebur,  working  in  our  laboratory,  recently  has  shown  that 
the  growth  of  very  young  embryos  of  common  barley  (0.3-1.1  mm.  long) 
may  be  greatly  improved  by  surrounding  them  on  a  nutrient  agar  medium 
with  aseptically  excised  endosperms.  The  basic  medium  employed  permits  a 
satisfactory  growth  of  older  barley  embryos  but  does  not  yield  transplantable 
seedlings  from  embryos  shorter  than  about  0.6  mm.  except  in  conjunction 
with  endosperms.  Coconut  milk  and  malt  extract  are  ineffective  with  barley 
embryos.  Water  extracts  of  fresh  barley  endosperms  gave  positive,  although 
smaller  effects  than  the  intact  tissue.  Further  exploration  of  the  living  endo- 
sperm as  a  source  of  nutrients  for  very  young,  excised  embryos  should  prove 
rewarding.  The  interrelationships  of  these  two  tissues  in  the  juvenile  seed 
give  strong  credence  to  this  approach.  The  success  which  has  so  often  at- 
tended efforts  to  grow  older  embryos  artificially  on  rather  simple  media  may 
have  blinded  us  to  the  fact  that  the  young  embryo,  divorced  from  the  endo- 
sperm, may  have  quite  different  requirements. 


W.  GORDON  WHALEY 

The  Plant  Research  Insfifufe  of  fhe  Universify  of  Texas  and 
the  Clayton  Foundation  for  Research 


Chapter  6 

Physiology  of  Gene 
Action  in  Hybrids 


The  physiology  of  gene  action  in  hybrids  is  not  a  subject  apart  from  the 
physiology  of  gene  action  in  organisms  in  general.  The  approach  to  specific 
problems  of  gene  action  is  probably  better  made  in  non-hybrid  organisms 
than  in  hybrids.  Hybrids  do,  however,  represent  one  type  of  genetic  situation 
which  in  certain  instances  is  particularly  favorable  for  the  study  of  gene 
action.  Most  useful  in  this  respect  are  those  hybrids  which  exhibit  the  phe- 
nomenon referred  to,  often  rather  loosely,  as  hybrid  vigor.  The  terms  hybrid 
vigor  and  heterosis  often  are  used  synonymously.  A  more  precise  usage,  and 
one  in  accord  with  the  original  definitions,  refers  to  the  developed  superior- 
ity of  hybrids  as  hybrid  vigor,  and  to  the  mechanism  by  which  the  superior- 
ity is  developed  as  heterosis.  By  this  definition,  hybrid  vigor  is  heterosis 
manifest.  Because  in  studies  of  growth  and  development  it  is  often  desirable 
to  distinguish  clearly  between  mechanism  and  end  result,  this  use  of  the  two 
terms  will  be  followed  in  this  chapter. 

Heterosis  has  been  the  subject  of  many  experiments  and  a  great  deal  of 
speculation  on  the  part  of  geneticists.  The  concern  has  been  mostly  with  the 
genetic  bases  of  heterosis,  and  relatively  little  attention  has  been  given  to  the 
physiological  mechanisms  involved.  As  a  matter  of  fact,  the  literature  on 
heterosis  mirrors  faithfully  the  changing  emphasis  in  genetics  in  the  last  two 
or  three  decades.  Practically  all  of  the  early  investigations  of  heterosis  had 
to  do  with  the  comparison  of  mature  characteristics  of  inbred  lines  and  their 
vigorous  hybrids,  and  then  with  attempts  to  formulate  genetic  schemes  in 
explanation  of  the  differences.  Gradually,  the  focus  of  investigation  has 
turned  to  a  study  of  developmental  differences  responsible  for  the  hybrid 
vigor,  and  more  recently  to  the  gene  action  bases  of  these  developmental 
differences. 

98 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  9^ 

It  is  a  fair  hope  that  from  detailed  studies  of  the  nature  and  development 
of  heterosis,  much  will  in  time  be  revealed  about  specific  gene  action.  Un- 
fortunately, most  of  the  studies  up  to  the  present  time  have  been  directed 
to  general  rather  than  to  specific  considerations.  It  has  been  necessary  to  deal 
in  terms  of  size  differences,  yield  differences,  and  growth  rate  differences,  un- 
til enough  of  the  pattern  should  appear  to  indicate  what  specific  physio- 
logical considerations  are  likely  to  be  involved  in  heterosis.  Because  we  have 
come  only  to  this  point  and  have  proceeded  but  a  little  way  in  an  analysis  of 
these  specific  physiological  considerations,  this  chapter  will  have  to  deal 
more  with  suggestions  of  the  likely  mechanisms  than  with  data  from  investi- 
gations of  them. 

It  is  neither  possible  nor  desirable  to  separate  wholly  the  consideration  of 
the  physiological  mechanisms  of  heterosis  from  the  genetic  bases.  Our  main 
concern  will  ultimately  be  with  the  genes  involved  and  the  nature  of  their 
action. 

The  word  hybrid  has  no  good,  definitive  genetic  meaning.  It  can  be  used 
with  equal  propriety  to  refer  to  organisms  which  approach  complete  hetero- 
zygosity or  to  organisms  which  are  heterozygous  for  only  a  small  number 
of  genes. 

There  is  at  least  a  rough  relationship  between  the  amount  of  heterosis  in  a 
hybrid  and  the  extent  of  the  genetic  differences  between  the  parents.  Physio- 
logical and  morphological  diversity  are  dependent  both  upon  the  number  of 
allelic  differences  between  organisms  and  upon  the  nature  of  the  action  of 
the  particular  genes  among  which  these  allelic  differences  exist.  It  is  quite 
possible  that  organisms  differing  by  only  a  few  genes  may  be  more  widely 
separated  in  certain  characteristics  than  are  organisms  differing  by  many 
more  genes — the  actions  of  which  are  of  less  fundamental  significance  for 
the  control  of  the  developmental  pattern. 

In  our  approach  to  questions  of  hybrid  vigor,  we  may  be  concerned  with 
different  degrees  of  hybridity.  Consideration  of  this  factor  must  involve  not 
only  the  number  of  genes  but  also  the  nature  of  the  action  of  the  particular 
genes.  Nor  is  this  all,  for  the  action  of  any  specific  allele  is  conditioned  by  the 
genetic  background  in  which  it  occurs  in  a  particular  individual.  Hence,  the 
relations  among  genes  may  often  be  of  critical  importance. 

Of  tremendous  import,  too,  are  the  interactions  between  the  activities  of 
the  genes  and  the  environment.  In  speaking  of  hybrid  vigor,  we  are  general- 
ly concerned  with  such  characteristics  as  size  and  yield,  but  these  are  merely 
end  products  of  the  metabolic  processes.  Patterns  of  these  metabolic  proc- 
esses are  set  by  the  genes,  but  the  processes  themselves  may  be  either  ac- 
celerated, inhibited,  or  otherwise  modified  by  the  effects  of  environmental 
factors.  Hybrids  which  are  particularly  vigorous  under  certain  conditions 
may  show  relatively  little  vigor  under  other  environmental  conditions.  It  is 
true  that  the  enhanced  vigor  of  hybrids  frequently  gives  to  them  a  wide 


TOO  W.  GORDON  WHALEY 

range  of  environmental  adaptability.  It  is  equally  true  that  certain  hybrids 
exhibit  vigor  within  only  relatively  narrow  environmental  limits.  For  lack  of 
evidence  it  must  be  assumed  that  the  distinction  lies  in  the  differences  be- 
tween the  patterns  of  hybridity  and  in  the  action  of  the  genes  responsible  for 
the  hybrid  advantages. 

Any  attempt  to  explain  the  genetic  basis  of  heterosis  must  make  initial 
recognition  of  one  fact.  The  phenomenon  can  involve  only  the  recombination 
of  alleles  already  existing  in  the  population  or  populations  from  which  the 
hybrid  organisms  have  been  developed;  unless,  by  rare  chance,  mutation 
should  take  place  just  prior  to  or  just  after  the  actual  crossing.  We  are  thus 
concerned  with  an  interpretation  limited  to  different  types  of  recombina- 
tions, and  to  different  kinds  of  gene  action  resulting  from  these  recombina- 
tions. 

GENETIC  MECHANISM  OF  HETEROSIS 

Consideration  of  the  characteristics  of  dominance  and  heterozygosity  has 
been  of  primary  importance  to  investigators  concerned  with  interpretation 
of  the  genetic  mechanism  of  heterosis.  Jones's  dominance  of  linked  factors 
hypothesis  (1917)  probably  is  still  the  most  popular  explanation  of  the 
genetic  basis  of  heterosis. 

Dobzhansky  (1941)  and  his  co-workers,  and  many  others,  have  recorded 
that  in  most  species  there  has  been,  in  the  course  of  evolution,  accumulation 
of  deleterious  recessive  characters,  which  when  homozygous  reduce  the 
efficiency  of  the  organism — but  which  in  the  heterozygous  condition  are 
without  efficiency-reducing  effects.  This  revelation  calls  for  a  reshaping  of  no- 
tions regarding  the  nature  of  the  favorable  effects  of  the  dominant  alleles,  but 
does  not  otherwise  modify  the  structure  of  the  explanation.  The  favorable- 
ness  of  the  action  of  many  of  the  dominant  alleles  probably  is  not  the  result 
either  of  directional  mutation  producing  more  favorable  dominants  or  of 
selection  tending  to  eliminate  the  unfavorable  dominants.  Instead,  it  may 
be  due  to  the  accumulation  in  populations  of  deleterious  recessive  mutations. 
These,  if  their  effects  are  not  too  deleterious,  often  can  be  piled  up  in  sig- 
nificant numbers. 

The  piling-up  of  such  deleterious  recessives  is  probably  one  of  the  reasons 
why  heterosis  is  a  much  more  important  phenomenon  in  such  a  plant  as  corn 
than  it  is,  for  example,  in  the  tomato.  Corn  has  been  handled  for  hundreds 
or  even  thousands  of  years  in  a  manner  that  has  made  possible  the  accumula- 
tion in  populations  of  relatively  large  numbers  of  deleterious  recessive  modi- 
fiers. The  tomato  is  more  than  90  per  cent  self-pollinated,  and  any  great 
accumulation  of  deleterious  modifiers  is  unlikely.  Corn  populations  char- 
acteristically contain  thousands  of  individuals,  and  wind  pollination  makes 
for  maintenance  of  heterozygosity.  In  tomato,  the  effective  breeding  popula- 
tion size  approaches  one,  and  deleterious  mutations  would  tend  to  become 


PHYSIOLOGY  OF  GENE  ACTION   IN   HYBRIDS  101 

homozygous  with  sufficient  frequency  to  bring  about  the  elimination  of 
many  of  them. 

As  a  matter  of  observation,  it  would  seem  that  a  comparison  of  the  occur- 
rence and  degree  of  heterosis  in  different  species,  along  with  a  consideration 
of  the  reproductive  mechanisms  in  the  various  species,  supports  the  proposal 
that  heterosis  in  many  cases  is  the  result  of  the  covering  uj)  in  the  hybrids  of 
deleterious  recessive  alleles  with  a  consequent  return  to  vigor.  The  often 
stated  argument  that  hybrids  of  corn,  for  instance,  frequently  are  more 
vigorous  than  the  original  open-pollinated  populations  from  which  the  in- 
breds  used  in  their  production  were  derived,  has  no  validity  with  respect  to 
this  situation.  In  the  production  of  the  inbreds  there  is  invariably  a  reassort- 
ing  of  the  alleles  of  the  open-pollinated  populations. 

It  is  highly  improbable,  however,  that  dominant  alleles  operating  either 
because  of  certain  inherent  favorable  characteristics  of  their  own,  or  simply 
to  prevent  the  deleterious  activity  of  recessives,  present  the  only  genetic 
basis  of  heterosis.  Dominance  is  by  no  means  the  clear-cut  feature  described 
in  Gregor  Mendel's  original  paper.  The  dominance  of  a  particular  allele  may 
be  conditioned  by  the  environment,  or  it  may  depend  upon  the  genetic 
background  in  which  the  allele  exists.  A  completely  dominant  effect  of  one 
allele  over  another,  in  the  classic  sense  of  our  utilization  of  the  word  domi- 
nance, is  by  no  means  universal. 

Rather  unfortunately  the  so-called  heterozygosity  concept  of  heterosis  has 
usually  been  introduced  as  being  in  opposition  to  the  dominance  explanation. 
Because  the  concepts  of  the  features  of  dominance  and  recessiveness  early 
put  them  into  rigid  categories,  it  has  been  difficult  to  postulate  how  a  hetero- 
zygous condition  with  respect  to  one  or  more  genes  could  render  an  organism 
more  vigorous  than  the  homozygous  condition,  usually  of  the  dominant 
alleles. 

Evidence  of  significance  for  the  interpretation  of  the  importance  of  hetero- 
zygosity in  heterosis  has  been  accumulated  slowly.  There  is  now,  however,  a 
fairly  long  list  of  instances  in  many  different  species  in  which  the  heterozy- 
gous condition  for  certain  alleles  is  known  to  be  superior  to  either  the  homo- 
zygous recessive  or  the  homozygous  dominant  condition  (Stubbe  and 
Pirshcle,  1940;  Singleton,  1943;  Karper,  1930;  Robertson,  1932;  Robertson 
and  Austin,  1935;  Gustafsson,  1938,  1946;  Nabours  and  Kingsley,  1934; 
Masing,  1938,  1939a,  1939b;  Rasmusson,  1927;  and  Timofeef-Ressovsky, 
1940. 

The  accumulation  of  data  on  these  cases  followed  a  long  period  during 
which  all  the  investigations  reported  seemed  to  indicate  no  marked  differ- 
ences between  organisms  heterozygous  for  certain  alleles  and  those  with  the 
dominant  homozygous  condition  for  these  same  alleles.  At  least,  in  no  in- 
stance, was  there  any  marked  superiority  referable  to  the  heterozygous 
condition.  Most  of  the  genes  involved  in  the  more  recent  findings  have  been 


102  W.  GORDON  WHALEY 

catalogued  as  having  at  least  moderately  deleterious  effects  in  the  mutated 
state.  The  characteristics  controlled  by  them  include:  chlorophyll  deficien- 
cies, modifications  of  leaf  form  and  pigmentation,  stalk  abnormalities,  flower- 
ing pattern,  and  time  of  flowering. 

The  extent  to  which  the  actual  nature  of  the  genetic  situation  has  been 
analyzed  varies,  but  in  several  of  the  cases  it  seems  clear  that  the  mutation 
of  a  single  gene  is  involved  and  that  the  Fi  hybrids  are  heterozygous  only 
with  respect  to  the  alleles  at  this  particular  locus.  The  amount  of  heterosis 
manifest  also  varies  greatly.  Because  of  experimental  differences,  no  accurate 
comparisons  can  be  made,  but  in  some  instances  the  amount  of  hybrid  vigor 
appears  to  be  nearly  comparable  to  that  which  occurs  in  crosses  involving 
large  numbers  of  allelic  differences.  The  situation  appears  to  be  one  in  which 
a  mutation  takes  place,  and  the  mutated  allele  is  definitely  deleterious  when 
homozygous.  In  individuals  heterozygous  for  the  particular  gene,  there  ap- 
pear none  of  the  deleterious  effects.  Instead,  a  definite  heterotic  effect  ap- 
pears. Dominance  is  of  no  apparent  importance,  and  the  distinction  between 
the  vigorous  hybrids  and  the  less  vigorous  non-hybrids  rests  upon  hetero- 
zygosity. 

Jones  (1944,  1945)  has  reported  several  cases  of  what  he  has  called  heter- 
osis resulting  from  degenerative  changes.  He  first  suggested  that  these  cases 
represented  instances  of  heterosis  with  a  genetic  basis  in  the  heterozygosity 
of  certain  of  the  mutated  genes.  More  recently  (private  communication) 
Jones  has  concluded  that  these  cases  involve  more  than  single  gene  differ- 
ences, and  that  the  results  may  be  explained  on  the  basis  of  an  accumulation 
of  favorable  dominant  effects. 

The  case  of  a  single  locus  heterosis  reported  by  Quinby  and  Karper  (1946) 
involves  alleles  which  do  not  produce  any  detectable  deleteriousness,  but  in 
certain  heterozygous  combinations  produce  hybrid  vigor  comparable  in 
amount  to  that  in  commercial  hybrid  corn.  Quinby  and  Karper  have  referred 
the  hybrid  advantage  in  this  case  to  a  stimulation  of  meristematic  growth  in 
the  heterozygous  plants. 

All  of  these  instances  involve  specific  allelic  interactions  and  not  superior- 
ity resulting  from  heterozygosity  per  se — as  was  postulated  by  some  of  the 
earlier  workers  concerned  with  the  genetic  interpretation  of  heterosis.  These 
examples  contribute  to  the  increasing  realization  that  the  phenomenon  of 
dominance  is  perhaps  of  less  importance  with  respect  to  heterosis  than  has 
been  supposed.  There  is  no  a  priori  reason  why  the  interaction  of  a  so-called 
recessive  allele  and  a  so-called  dominant  allele  should  not  give  results  differ- 
ent from  and  metabolically  superior  to  those  which  are  conditioned  by  either 
two  recessives  or  two  dominants. 

This  situation  bears  closely  upon  the  interpretation  of  heterosis  set  forth 
by  East  in  1936.  East  postulated  that  at  the  loci  concerned  with  the 
mechanism  of  heterosis  there  might  be  a  series  of  multiple  alleles — with  the 
combinations  of  different  alleles  giving  results  metabolically  superior  to 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  103 

those  determined  by  the  combinations  of  like  alleles,  and  with  no  considera- 
tions of  dominance  being  involved.  In  the  light  of  existing  evidence  it  seems 
a  safe  assumption  that  a  considerable  portion  of  hybrid  vigor  is  the  result 
of  allelic  interaction  between  different  alleles  at  the  same  locus.  Although  the 
evidence  as  yet  is  scanty,  it  is  certainly  pertinent  to  suggest  that  some 
heterosis  may  result  from  the  interaction  of  alleles  at  different  loci,  when 
such  alleles  are  brought  into  new  combinations  in  the  hybrids. 

Most  of  the  recent  studies  of  the  relation  of  heterozygosity  to  heterosis 
have  been  concerned  with  the  results  of  the  action  of  single  genes.  Such 
studies  have  emphasized  that  heterosis  need  not  have  its  basis  in  the  action 
of  large  numbers  of  genes  but  can  be,  and  apparently  frequently  is,  a  result 
of  the  combining  of  different  alleles  of  a  single  gene.  Any  considerable  amount 
of  hybrid  vigor  resulting  from  the  action  of  single  genes  would  seem  to  indi- 
cate the  involvement  either  of  multiple  effects  of  single  genes  or  of  genie 
action  in  the  control  of  relatively  fundamental  metabolic  processes.  Both  are 
likely  probabilities. 

The  metabolic  system  of  any  organism  which  grows  and  functions  in  a 
satisfactory  manner  is  an  exceedingly  complicated  mechanism  with  a  great 
number  of  carefully  balanced,  interrelated  processes.  The  mutation  of  any 
gene  which  has  control  over  any  of  the  key  processes  or  functions  will  almost 
certainly  be  reflected  in  a  number  of  processes  and  activities.  For  example,  if 
a  change  in  the  character  of  some  fundamental  enzyme  system  is  involved, 
either  the  addition  or  subtraction  of  a  functional  step,  or  of  a  substance 
produced  at  a  particular  developmental  stage,  would  be  likely  to  enhance  or 
inhibit  a  number  of  important  processes  in  the  general  metabolism  of  the 
organism. 

The  equilibrium  factor  in  genie  action  is  obviously  a  consideration  of 
great  importance.  If  a  mutation  disturbs  this  equilibrium  after  it  has  become 
fairly  well  established  through  selection  and  elimination  processes,  the  con- 
sequences may  reduce  the  organism's  vigor.  If,  in  a  hybrid,  the  mutation  is 
then  brought  together  with  the  original  wild  type  or  normal  allele,  the  sum 
total  of  the  actions  of  the  mutated  allele  and  the  original  allele  may  well  be 
such  as  to  exceed  that  of  two  copies  of  the  original  allele  in  the  production 
of  vigor  in  the  organism. 

When  we  give  attention  to  physiology  of  gene  action  in  hybrids  which  are 
heterotic,  we  must  concern  ourselves  with  all  of  these  considerations  in- 
cluding the  fact  that  a  single  gene,  the  mutation  of  which  affects  some 
processes  in  a  sufficiently  fundamental  stage  of  the  organism's  formation, 
may  well  have  a  greater  end  effect  than  a  number  of  genes  whose  functions 
are  concerned  with  more  superficial  developmental  processes. 

SEED  AND  EMBRYO  DEVELOPMENT 

The  literature  on  heterosis  contains  a  number  of  discussions  concerning 
the  relation  between  seed  and  embryo  size  and  heterosis  (Kiesselbach,  1926: 


104  W.  GORDON  WHALEY 

Ashby,  1930,  1932,  1937;  East,  1936;  Sprague,  1936;  Copeland,  1940;  Mur- 
doch, 1940;  Kempton  and  McLane,  1942;  Whaley,  1944,  1950). 

Most  of  the  investigations  have  dealt  with  mature  seed  and  embryo  size. 
The  evidence  shows  that  in  many  instances  hybrid  vigor  is  associated  with  a 
high  embryo  weight.  In  some  cases  the  initially  high-weight  embryo  is  found 
in  a  relatively  large  seed.  There  is,  however,  by  no  means  a  consistent  correla- 
tion between  either  high  embryo  weight  or  large  seed  size  and  heterosis. 

The  results  of  studies  on  corn  inbreds  and  hybrids  in  our  own  laboratory 
(Whaley,  1950)  seem  representative  of  the  general  findings.  Among  some  ten 
inbred  lines  there  occurred  a  great  deal  of  variation  from  one  line  to  another 
as  to  both  embryo  weight  and  seed  weight.  There  was  somewhat  more  varia- 
tion with  respect  to  embryo  weight.  Among  the  Fi  hybrids,  all  of  which 
exhibited  considerable  vigor  under  central  Texas  conditions,  there  were  a 
few  with  embryo  weights  which  exceeded  those  of  the  larger-embryo  parent. 
For  the  most  part,  the  embryo  weights  were  intermediate,  and  in  one  or  two 
cases  they  were  as  low  as  that  of  the  smaller-embryo  parent.  The  weight  of 
the  seed  tissues  other  than  the  embryo  tended  to  follow  that  of  the  pistillate 
parent,  but  was  generally  somewhat  higher.  Double  crosses  which  had  vigor- 
ous Fi  hybrids  as  pistillate  parents  characteristically  had  large  seeds  with 
what  were  classified  as  medium-weight  embryos. 

The  few  reports,  such  as  Copeland's  (1940),  concerning  the  development 
of  embryos  in  inbred  and  hybrid  corn,  suggest  that  at  the  earlier  stages  of 
development  some  hybrid  vigor  is  apparent  in  the  hybrid  embryos.  The 
observations  of  hybrid  vigor  during  early  development  of  embryos  and  the 
absence  of  any  size  advantage  at  the  time  of  seed  maturity  are  not  necessari- 
ly conflicting.  In  most  plants,  embryo  and  seed  maturation  represent  fairly 
definite  stages  at  which  a  certain  degree  of  physiological  maturity  and  of 
structural  development  has  been  attained.  It  is  probably  to  be  anticipated 
that  even  though  certain  heterotic  hybrids  show  early  embryo  development 
advantages,  these  advantages  may  be  ironed  out  by  the  time  the  embryo 
and  the  seed  mature.  The  size  of  both  the  embryo  and  the  other  seed  tissues 
is  conditioned  not  only  by  the  genotype  of  these  tissues  themselves,  but  also 
by  the  nutritional  background  furnished  them  by  the  plant  on  which  they 
grow. 

It  is  quite  possible  that  this  genotype-to-background  relationship  is  an 
important  consideration  in  the  determination  of  whether  or  not  hybrid  vigor 
is  exhibited  in  the  development  of  the  embryo  and  seed.  The  background 
provided  by  the  pistillate  parent  might  be  such  as  to  preclude  the  develop- 
ment of  embryo  vigor,  even  though  the  embryo  genotype  were  of  a  definitely 
heterotic  constitution.  The  fact  that  hybrid  vigor  is  apparent  during  certain 
stages  of  embryo  and  seed  development  may  or  may  not  be  related  to  an 
embryo  or  seed  size  advantage  at  maturity.  Because  of  this,  it  seems  doubt- 
ful that  embryo  or  seed  size  is  a  reliable  measure  of  hybrid  vigor;  and  that 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  105 

the  rate  of  development  during  the  embryo  and  seed  maturation  i)eriod  is 
of  any  critical  importance  with  respect  to  the  development  of  hybrid  vigor 
during  post-embryonic  growth. 

EARLY  SEEDLING  GROWTH  AND  HETEROSIS 

There  have  been  few  studies  of  early  postgermination  growth  in  plants  in 
relation  to  heterosis.  It  would  seem  that  the  usual  failure  to  find  higher 
growth  rates  during  the  grand  period  of  growth,  or  longer  continued  growth 
periods  in  heterotic  hybrids,  would  suggest  that  the  answer  to  the  develop- 
ment of  hybrid  vigor  lies  for  the  most  part  in  the  early  postgermination 
growth  stages.  The  work  of  Ashby  and  his  co-workers  (Ashby,  1930,  1932, 
1936;  Hatcher,  1939, 1940;  Luckwill,  1937, 1939)  emphasized  that  the  hybrid 
advantage  in  their  materials  was  either  present  in  the  resting  embryo  or  be- 
came manifest  in  early  postgermination  growth.  Its  development  was  defi- 
nitely not  a  characteristic  of  the  later  growth  phases.  This  observation  has 
now  been  made  for  many  cases  of  hybrid  vigor  (Whaley,  1950).  There  are 
some  instances  in  which  hybrid  vigor  seems  to  be  the  result  of  longer-con- 
tinued growth  on  the  part  of  the  hybrid.  These  probably  have  a  dilTerent 
explanation  from  the  majority  of  cases. 

We  have  been  concerned  lately  in  our  own  laboratory  with  an  analysis 
of  the  early  postgermination  growth  of  corn  inbreds  and  single  and  double 
cross  hybrids  (Whaley,  1950) .  Studies  of  growth  during  the  first  ten  to  twelve 
days  after  germination  have  revealed  that  the  hybrid  advantage  is  largely 
the  result  of  the  heterotic  hybrid  plants  reaching  a  high  growth  rate  earlier 
than  do  the  inbreds.  Almost  without  exception,  the  development  of  the  hy- 
brid advantage  takes  place  very  rapidly  in  the  early  stages  of  germination 
and  growth.  Rarely  have  we  seen  evidence  of  the  hybrids  having  higher 
growth  rates  during  any  later  part  of  the  developmental  cycle.  Neither  are 
the  hybrid  growth  periods  extended  appreciably  beyond  those  of  the  in- 
breds. In  most  instances  the  hybrids  mature  somewhat  more  rapidly  than 
the  inbreds — a  fact  of  common  observation  among  plant  breeders. 

Since  the  attainment  of  the  maximum  growth  rate  takes  place  more 
quickly  during  the  early  stages  of  development,  the  hybrids  do  have  a  longer 
maximum  growth  rate  period.  During  this  period  the  early  advantage  is 
compounded,  to  give  a  considerably  greater  maturity  advantage.  When 
both  the  inbred  lines  and  the  hybrids  used  in  our  studies  are  considered,  it  is 
apparent  that  the  rapid  attainment  of  high  early  growth  rates  is  correlated 
with  relatively  low  embryo  weights.  This  apparent  higher  efficiency  of  small 
embryos  and  its  importance  in  relation  to  hybrid  vigor  requires  further  study. 

On  the  basis  of  the  data  at  hand  one  can  suggest  that  the  hybrid  advantage 
lies  in  the  more  rapid  unfolding  of  certain  metabolic  processes,  a  suggestion 
which  receives  support  from  the  recorded  studies  of  later  growth. 


106  W.  GORDON  WHALEY 

LATER  GROWTH  AND  HETEROSIS 

It  is  unfortunate  that  most  studies  of  the  physiology  of  heterosis  have  been 
confined  to  the  later  growth  period,  and  consequently  do  not  include  that 
part  of  the  growth  cycle  during  which  the  important  differences  seem  to  be 
developed.  Nonetheless,  we  can  learn  much  from  these  studies  of  later 
growth  as  to  the  nature  of  the  physiological  differences  which  may  furnish 
bases  for  the  development  of  hybrid  vigor. 

The  early  experiments  on  physiological  differences  between  inbreds  and 
hybrids  were  concerned  mostly  with  the  responses  of  the  inbreds  and  the 
hybrids  to  different  soil  conditions.  A  few  examples  will  serve  to  indicate  the 
type  of  investigation  and  the  character  of  the  results.  Hoffer  (1926)  deter- 
mined the  amounts  of  the  constituents  of  the  ash  of  heterotic  hybrid  corn  to 
be  generally  intermediate  between  those  of  the  parental  types.  He  noted  that 
iron  and  aluminum  were  present  in  the  ash  of  the  hybrids  in  smaller  amounts 
than  in  the  inbreds.  His  studies  showed  that  although  there  were  marked 
differences  in  the  absorption  of  iron  and  aluminum  in  different  soil  types  the 
vigorous  hybrids  tended  to  absorb  less  of  both  these  elements  than  the  less 
vigorous  inbred  lines. 

In  the  same  year  Kiesselbach  (1926)  reported  distinct  differences  in  water 
requirements  between  selfed  lines  of  corn  and  their  heterotic  F;  hybrids.  The 
low  productivity  inbreds  had  much  higher  water  requirements  than  the 
vigorous  Fi  hybrids,  when  water  requirements  were  calculated  on  the  basis  of 
either  water  absorbed  per  gram  of  ear  corn  or  water  absorbed  per  gram  of 
total  dry  matter.  Barley  inbreds  and  heterotic  barley  hybrids  were  shown 
by  Gregory  and  Crowther  (1928,  1931)  to  make  distinctly  different  responses 
to  various  levels  of  available  minerals.  These  investigators  postulated  that 
heterosis  in  barley  might  be  directly  related  to  differences  in  the  ability  of 
the  hybrids  and  the  inbreds  to  use  certain  nutrients.  This  suggestion  has  had 
a  fairly  adequate  test,  particularly  with  reference  to  nitrogen  and  phos- 
phorus nutrition. 

The  work  of  DeTurk  et  al.  (1933),  Smith  (1934),  Lyness  (1936),  Harvey 
(1939),  Burkholder  and  McVeigh  (1940),  and  Rabideau  et  al.  (1950),  has 
provided  a  fairly  adequate  picture  of  the  relation  of  phosphorus  and  nitro- 
gen nutrition  to  the  development  of  hybrid  vigor.  Smith  demonstrated  dis- 
tinct differences  among  inbred  corn  lines  with  respect  to  phosphorus  nutri- 
tion, noting  that  these  differences  were  most  apparent  when  the  phosphorus 
supply  was  limited.  He  postulated  that  the  higher  phosphate  utilization  effi- 
ciency of  the  hybrids  might  be  referred,  at  least  in  part,  to  the  dominant  in- 
heritance in  them  of  a  much  branched  root  system.  Later  studies  have  shown 
that  the  root  growth  pattern  is  certainly  important  in  relation  to  heterosis. 

Smith  noted  particularly  that  when  inbred  lines  were  inefficient  in  the 
utilization  of  phosphorus  or  nitrogen,  crossing  them  failed  to  produce  hybrids 
showing  any  evidence  of  physiological  stimulation  resulting  in  the  more 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  107 

efTective  use  of  these  elements.  Lyness  (1936)  studied  heterotic  Fi  hybrids 
resulting  from  crosses  between  a  low  phosphorus-absorbing  capacity  inbred 
and  a  high  phosphorus-absorbing  capacity  inbred.  He  found  the  heterotic 
Fi  plants  to  have  high  phosphorus-absorbing  capacity.  These  results  sug- 
gested that  phosphorus-absorbing  capacity  in  corn,  in  some  instances  at 
least,  acts  genetically  as  a  dominant  factor.  Lyness  also  noted  the  relation- 
ship between  high  phosphorus  absorption  and  the  extent  of  root  develop- 
ment. He  supposed  that  the  extent  of  root  development  might  be  responsible 
for  varietal  differences  in  phosphorus  absorption,  a  supposition  which  is  sup- 
ported by  later  studies.  The  work  of  DeTurk  et  at.  (1933)  suggested  that  more 
than  simply  phosphorus-absorbing  capacity  is  involved.  This  work  revealed 
that  the  actual  phosphorus  content  patterns  of  two  Fi  hybrids  of  corn  were 
quite  different.  By  estimating  the  amount  of  phosphorus  in  various  chemical 
fractions,  De  Turk  and  his  coworkers  were  able  to  demonstrate  marked  phos- 
])horus  pattern  difTerences  and  to  associate  these  pattern  differences  with 
various  phosphate  fertilizer  treatments. 

In  our  laboratory  we  have  made  a  study  of  the  phosphorus-absorbing  ef- 
ficiency of  corn  inbreds  and  hybrids,  and  have  attempted  to  correlate  the 
findings  of  this  study  with  developmental  changes  in  the  vascular  system 
and  with  general  growth  (Whaley  et  al.,  1950;  Heimsch  et  al.,  1950;  Rabideau 
et  al.,  1950).  The  data  indicate  that  heterotic  hybrids  definitely  absorb  more 
radioactive  phosphorus  than  their  inbred  parents.  This  advantage  in  ab- 
sorption on  the  part  of  the  hybrid  is  associated  with  more  rapid  early  de- 
velopment, with  earlier  attainment  of  maturity,  and  with  certain  features  of 
vascular  organization.  The  greater  absorption  can  be  referred  at  least  in 
part  to  better  early  development  of  the  root  system  in  the  hybrids,  and  to  a 
generally  higher  level  of  metabolic  activity  which  presumably  creates  a 
greater  phosphorus  demand.  The  greater  absorption  of  phosphorus  by  the 
hybrids  is  certainly  one  of  the  factors  which  compounds  the  heterotic  effects, 
but  it  seems  doubtful  that  it  is  a  primary  factor  in  the  development  of  hybrid 
vigor. 

Harvey's  (1939)  studies  of  nitrogen  metabolism  among  inbreds  and  hy- 
brids of  both  corn  and  tomato  showed  differences  from  one  line  to  another 
with  respect  to  the  ability  to  use  nitrate  and  ammonium  nitrogen.  The  ex- 
periments were  of  such  a  nature  as  to  make  it  clear  that  such  differences  in 
nutritional  responses  were  results  of  differences  in  genetic  constitution. The 
behavior  of  hybrids  produced  from  the  inbreds  reflected  a  combination  of  the 
characteristics  of  the  inbreds.  Significantly,  Harvey's  study  revealed  that  not 
only  did  differences  exist  among  his  inbreds  and  hybrids  with  respect  to  the 
ability  to  use  different  types  of  nitrogen,  but  that  there  were  distinct  genetic 
differences  in  the  responses  of  the  plants  to  various  levels  of  nitrogen  avail- 
ability. 

Somewhat  similar  differential  responses  to  potassium  availability  were 


108  W.  GORDON  WHALEY 

revealed  by  Harvey's  studies  on  tomato  inbreds  and  hybrids.  Burkholder  and 
McVeigh  (1940)  have  also  noted  differences  in  responses  of  corn  inbreds  and 
hybrids  to  various  levels  of  available  nitrogen.  These  investigators  corre- 
lated apical  meristematic  development,  and  the  differentiation  of  the  vascu- 
lar system  with  the  level  of  nitrogen  nutrition,  and  the  efficiency  of  different 
lines  and  hybrids  in  utilizing  the  available  nitrogen.  Their  results  indicate 
that  hybrid  vigor,  involving  superiority  in  the  production  of  dry  matter 
and  the  differentiation  of  organs,  was  not  correlated  with  greater  growth  and 
development  of  the  vascular  system. 

There  definitely  are  vascular  organization  differences  between  the  heterot- 
ic  hybrids  and  the  inbreds  in  the  material  we  have  studied.  These  vascular 
organization  differences  seem  not  to  be  the  result  of  differences  in  mineral 
absorption  and  distribution,  but  rather  to  be  one  of  the  factors  responsible 
for  the  differences  in  absorption  and  distribution.  All  the  evidence  seems  to 
indicate  that  the  greater  absorption  of  minerals  by  heterotic  hybrids  can  be 
referred  to  better  developed  root  systems  in  the  hybrids,  probably  also  to  the 
presence  of  more  efficient  transport  systems,  and  to  a  generally  higher  level 
of  metabolic  activity. 

Recently  we  have  undertaken  a  rather  extensive  analysis  of  both  the 
morphological  and  physiological  characteristics  of  a  tomato  cross  in  which 
there  is  marked  heterosis.  We  have  found  no  significant  differences  between 
the  inbreds  and  the  hybrids  as  to  total  phosphorus  content  of  the  leaves, 
stems,  or  roots.  There  is  some  suggestion  that  the  phosphorus  content  of  the 
organs  of  the  hybrids  reaches  a  higher  level  earlier  in  growth  than  it  does  in 
the  inbreds.  Neither  do  the  hybrid  plants  have  any  consistent  advantage 
with  respect  to  nitrogen  content. 

Analyses  of  the  starch  content  of  the  leaves  and  stems  suggest  that  the 
hybrid  plants  may  have  a  slightly  higher  starch  content  than  the  inbreds 
during  the  early  growth  stages.  In  terms  of  average  figures  over  the  whole 
growth  period,  however,  there  are  no  marked  differences  between  the  in- 
breds and  the  hybrids.  The  same  appears  to  be  true  of  the  sugar  content. 
The  hybrids  have  a  somewhat  higher  sugar  content,  at  least  in  the  leaves, 
early  in  development.  During  the  greater  part  of  the  growth  cycle  the  hy- 
brids do  not  have  significantly  more  sugar  than  the  inbreds.  The  only  clear 
difference  found  between  the  inbreds  and  the  hybrids  is  in  the  catalase  ac- 
tivity of  the  shoot  tips,  the  hybrids  having  an  appreciably  greater  index  of 
catalase  activity  than  either  of  the  inbred  parents.  The  catalase  activity 
differences  are  associated  with  much  more  active  meristematic  growth  in  the 
hybrid  plants. 

THE  ROLE  OF  SPECIFIC  SUBSTANCES  IN  HETEROSIS 

Evidence  for  another  sort  of  physiological  differences  possibly  involved  in 
heterosis  is  furnished  by  the  work  of  Robbins  (1940,  1941a)  in  assaying  the 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  109 

growth-promoting  activities  of  extracts  from  inbred  and  hybrid  corn  grains. 
Robbins'  evidence  indicates  that  a  substance  or  substances,  which  he  has 
designated  as  factor  Z,  may  be  synthesized  in  greater  amounts  by  the  hy- 
brids than  by  the  inbreds.  He  has  stated  that  factor  Z  can  be  fractionated  into 
Zi,  which  is  hypoxanthine;  and  Z2,  a  still  unidentified  fraction.  Robbins' 
work  suggests  that  among  the  advantages  possessed  by  heterotic  hybrids 
may  be  the  ability  to  synthesize  certain  growth  substances  which  the  in- 
breds either  cannot  synthesize  or  cannot  synthesize  as  well. 

Further  evidence  of  a  slightly  different  nature  is  provided  by  the  root 
culture  work  of  Robbins  (1941b)  and  of  Whaley  and  Long  (1944).  Robbins 
used  cultures  of  a  strain  of  Lycopersicou  pimpiiieUijolium  Mill.,  a  strain  of 
L.  esculeutum  Mill.,  and  their  Fi  hybrid,  in  solutions  supplemented  by  thia- 
min, thiamin  and  pyridoxine,  or  thiamin,  pyridoxine,  and  nicotinamide. 
Robbins  found  that  the  Fi  roots  grew  much  more  rapidly  and  produced 
more  dry  matter  than  those  of  either  parental  line.  He  was  able  to  show 
further  that  one  parental  line  made  a  greater  response  to  the  presence 
of  pyridoxine  than  did  the  other,  while  the  roots  of  the  second  parental  line 
made  a  greater  response  to  nicotinamide  than  those  of  the  first.  This  suggests 
the  combination  of  complementary  factors  from  the  parents  in  the  hybrid. 
Whaley  and  Long  (1944)  obtained  essentially  the  same  results  with  a  cross 
involving  two  inbred  lines  of  L.  esculeutum. 

In  the  University  of  Texas  tissue  and  organ  culture  laboratory,  we  have 
been  exploring  certain  aspects  of  this  problem.  While  the  results  are  not  suf- 
ficiently complete  for  publication,  some  facts  are  already  clear.  Among  the 
roots  of  many  inbred  lines  of  tomatoes  which  we  have  been  culturing,  there 
are  marked  differences  in  growth  responses  associated  with  the  availability 
or  non-availability  of  thiamin,  pyridoxine,  niacin,  and  certain  other  sub- 
stances. These  differences  appear  definitely  to  be  inherited  and  they  can  be 
studied  in  either  the  inbred  lines  or  hybrids. 

It  is  still  too  early  to  say  what  the  inheritance  pattern  is,  but  consideration 
can  be  given  to  some  aspects  of  the  growth  response  patterns.  One  of  the 
most  significant  revelations  is  that  the  responses  of  most  of  the  roots  to  a 
specific  substance  are  conditioned  not  only  by  the  availability  of  that  sub- 
stance, but  by  the  availability  of  the  other  substances  and  by  the  gen- 
eral composition  of  the  culture  medium.  Heterosis  in  tomato  root  cultures 
is,  like  heterosis  in  whole  plants,  definitely  relative,  and  conditioned,  not 
only  by  the  environment,  but,  with  respect  to  any  specific  gene  action,  by 
the  background  of  other  gene  actions  taking  place  in  the  developing  or- 
ganism. 

Heterosis  in  tomato  root  cultures  is  definitely  related  to  the  inheritance  of 
the  capacity  to  synthesize  or  utilize  such  substances  as  thiamin,  ])yridoxine, 
and  niacin.  This  is  not  to  suggest  that  heterosis  in  whole  plants  of  tomato 
may  have  its  basis  in  the  genetic  recombination  of  factors  concerned  in  the 


no  W.  GORDON  WHALEY 

control  of  thiamin,  pyridoxine,  or  niacin  metabolism.  In  intact  plants,  it  is 
likely  that  the  green  parts  supply  these  substances  to  their  own  tissues  and  to 
the  roots,  in  amounts  satisfactory  for  growth  and  development.  The  root 
tissue  responses,  however,  are  definitely  heterotic  in  certain  instances,  and 
these  mechanisms  merit  examination. 

It  seems  pertinent  to  explore  the  role  of  these  B  vitamins  in  growth  and 
development.  Thiamin  appears  to  be  a  metabolic  requirement  for  all  types  of 
cells.  Its  metabolic  activity  apparently  revolves  around  a  role  in  enzyme 
systems.  Thiamin  pyrophosphate  is  the  co-enzyme  of  the  enzyme  pyruvate 
carboxylase  (Lohmann  and  Schuster,  1937).  The  enzyme  carboxylase  occurs 
in  many  plant  tissues.  The  possible  biochemical  basis  of  thiamin  action  in 
plants  has  been  set  forth  in  some  detail  by  Bonner  and  Wildman  (1946), 
Vennesland  and  Felsher  (1946),  and  Bonner  and  Bonner  (1948).  It  is  assumed 
that  thiamin  represents  a  step  in  the  development  of  co-carboxylase  which  is 
active  in  one  or  more  of  the  decarboxylating  enzyme  systems  of  the  respira- 
tory mechanism. 

Pyridoxine  also  has  an  enzymatic  role,  apparently  being  important  for  its 
conversion  to  pyridoxal  phosphate,  which  is  a  co-enzyme  of  one  or  more  of 
the  reactions  in  the  nitrogen  metabolism  of  the  plant  (Bonner  and  Bonner, 
1948).  As  a  co-enzyme  active  in  nitrogen  metabolism  reactions,  pyridoxine 
may  be  of  extreme  importance  in  amino  acid-protein  building,  and  hence 
active  in  conditioning  fundamental  growth  activities. 

Similarly,  niacin  activity  is  enzymatic  in  character.  Niacin  appears  to  be 
involved  as  a  constituent  of  the  nucleotide  cozymase,  and  possibly  of  tri- 
phosphopyridine  nucleotide.  Cozymase  is  a  co-enzyme  for  a  whole  series  of 
dehydrogenase  enzymes,  including  alcohol  dehydrogenase,  malic  dehydrog- 
enase, and  glutamic  dehydrogenase  (Bonner  and  Bonner,  1948). 

The  genetic  background  of  thiamin,  pyridoxine,  and  niacin  metabolism  is 
thus  a  genetic  background  concerned  with  basic  components  of  the  plant's 
enzyme  systems.  Heterosis,  which  rests  upon  recombinations  concerned  with 
thiamin,  pyridoxine,  or  niacin  metabolism,  quite  obviously  rests  upon  recom- 
binations which  are  concerned  with  the  acceleration,  inhibition,  or  blocking 
of  specific  stages  or  developed  substances  in  the  basic  enzyme  system. 

A  considerable  amount  of  supporting  evidence  for  the  involvement  of  such 
fundamental  enzyme  and  other  growth  substance  activities  in  the  develop- 
ment of  heterosis  has  been  coming  for  some  time  from  the  work  on  Neuro- 
spora.  In  many  heterocaryons  of  Neurospora,  increased  growth  responses 
directly  suggestive  of  heterosis  have  been  observed.  In  a  number  of  instances 
(Beadle  and  Coonradt,  1944),  the  growth  responses  depend  upon  the  two 
types  of  nuclei  in  the  heterocaryon — each  carrying  wild  type  alleles  of  de- 
leterious mutant  genes  carried  by  the  other  nucleus.  Such  instances  represent 
essentially  the  same  situation  as  the  recombination  of  favorable  dominant 
alleles  in  normally  diploid  organisms. 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  111 

In  one  case  reported  by  Emerson  (1948)  a  different  situation  obtains.  A 
mutant  strain  of  Neurospora  which  requires  sulfonamides  for  growth  at  cer- 
tain temperatures  will  grow  satisfactorily  in  the  absence  of  sulfonamides, 
provided  that  the  concentration  of  available  p-aminobenzoic  acid  is  held  at  a 
particular  level.  Either  higher  or  lower  concentrations  of  p-aminobenzoic  acid 
result  in  growth  inhibitions.  Emerson  has  made  heterocaryons  between  a 
mutant  strain  carrying  the  sulfonamide-requiring  gene  {sfo)  and  a  gene  which 
prevents  the  synthesis  of  p-aminobenzoic  acid  (pab),  and  a  strain  carrying 
sfo  and  the  wild  type  allele  (+)  of  pab.  The  resultant  heterocaryons  grow 
vigorously  on  the  minimal  medium  (without  sulfonamides),  whereas  strains 
carrying  sfo  and  pab,  or  sfo  and  +,  make  no  appreciable  growth  on  the 
minimal  medium.  Emerson's  explanation  of  the  growth  of  the  hetero- 
caryons is  that  it  results  from  a  balance  between  the  production  of  p-amino- 
benzoic acid  by  one  of  the  types  of  nuclei  and  the  absence  of  production  of 
p-aminobenzoic  acid  by  the  other  type  of  nucleus;  so  that  the  total  produc- 
tion of  p-aminobenzoic  acid  is  sufBcient  for  growth  but  still  within  the  range 
tolerated  by  strains  carrying  sfo.  Heterosis-like  effects  of  this  sort  are  sugges- 
tive of  the  instances  of  heterosis  related  to  the  heterozygosity  of  particular 
genes  in  diploid  organisms. 

We  thus  have  in  Neurospora,  heterosis-like  effects  assignable  both  to  a 
recombination  of  dominant  alleles  basis  and  to  a  heterozygosity  basis.  More 
important  for  this  discussion  is  the  fact  that  these  instances  are  all  concerned 
with  facilitation  in  the  hybrid  of  the  production  or  utilization  of  substances 
which  are  components  of  the  basic  enzyme  or  other  growth  substance  pat- 
tern of  the  organisms. 

Various  investigations  of  heterosis  in  Drosophila,  while  for  the  most  part 
not  concerned  with  specific  growth  substances,  have  nonetheless  assigned 
manifestation  of  heterosis  to  a  background  in  the  fundamental  biochemical 
activities  of  the  organisms.  Inasmuch  as  these  investigations  are  discussed  in 
detail  in  another  chapter,  they  will  not  be  treated  here. 

THE  PHYSIOLOGICAL  BASIS  OF  HETEROSIS 

From  consideration  of  the  pertinent  data,  a  definite  pattern  emerges. 
This  associates  the  development  of  heterosis  with  the  ability  of  the  hybrid 
to  synthesize  or  to  utilize  one  or  several  specific  substances  involved  in  the 
fundamental  growth  processes  of  the  organisms.  Nutritional  factors,  water 
absorption  factors,  and  the  other  more  gross  considerations  with  which  in- 
vestigators have  been  particularly  concerned  seem  to  be  secondary  factors^ — 
perhaps  responsible  for  compounding  the  heterotic  effects  but  probably  not 
responsible  for  their  initial  development.  Much  of  the  evidence  agrees  with 
the  assumption  that  the  primary  heterotic  effect  is  concerned  with  growth 
substances  whose  predominant  activity  is  registered  in  the  early  part  of 
the  developmental  cycle;  in  plants,  especially  in  early  postgermination 


112  W.  GORDON  WHALEY 

growth.  Into  this  category  fall  the  enzymes,  the  auxins,  and  the  other  "phys- 
iological key"  substances. 

Many  heterotic  hybrid  plants  seem  to  gain  their  advantage  within  the  first 
few  hours  after  germination.  This  advantage  may  not  be  shown  as  statistical- 
ly significant  until  it  has  been  further  heightened  by  subsequent  growth. 
The  primary  growth  activities  during  this  period  are  those  involved  in  the 
unfolding  of  the  enzymatic  pattern;  the  mobilization,  transformation,  and 
utilization  of  stored  materials,  and  the  building  up  of  active  protoplasmic 
synthesis.  It  seems  definitely  to  be  here  that  the  hybrid  advantage  lies.  By 
the  time  growth  is  well  under  way,  the  hybrid  advantage  is  already  well 
developed. 

Structural  diflferences  between  inbreds  and  heterotic  hybrids  shown  by  the 
studies  of  Burkholder  and  McVeigh  (1940),  Weaver  (1946),  and  the  members 
of  our  laboratory  (Whaley  et  al.,  1950;  Heimsch  et  al.,  1950;  Rabideau  ef  al., 
1950)  are  apparently  to  be  regarded  as  results  of  heterosis  rather  than  as 
causal  factors.  The  evidence  suggests  that  heterosis  is  concerned  primarily 
with  growth  processes  and  that  differentiation  activities  are  most  likely  in- 
volved secondarily  rather  than  primarily.  What  seems  to  be  indicated  is  the 
assignment  of  the  physiological  basis  of  heterosis  to  the  activity  of  one  or 
more  of  the  so-called  physiologically  active  substances  involved  in  early 
growth. 

Much  of  the  apparent  hybrid  vigor  is  assignable  to  these  activities  only  in 
a  secondary  fashion.  Once  the  advantage  of  a  larger  number  of  growing 
centers  or  of  heightened  meristematic  activity  is  established,  the  greater 
availability  of  nutrients,  the  greater  amount  of  protoplasm  involved  in 
further  protoplasm  building,  and  other  general  advantages  tend  to  increase 
the  initial  differences.  To  the  general  evidence  in  favor  of  this  supposition 
can  be  added  the  specific  evidence  of  the  few  cases  in  which  the  physiological 
action  of  particular  alleles  is  known.  Where  these  alleles  in  combination  are 
responsible  for  heterosis,  they  have — when  studied  in  sufficient  detail — 
invariably  been  shown  to  be  alleles  whose  action  involves  basic  enzyme  or 
other  growth  substance  activity. 

If  we  are  to  make  significant  headway  in  understanding  the  physiological 
mechanism  of  heterosis,  we  shall  have  to  concentrate  on  a  detailed  study  of 
the  developmental  physiology  of  early  growth.  Much  of  the  general  knowl- 
edge we  already  have  can  contribute  toward  this  understanding  if  we  trans- 
late it  into  terms  signifying  that  when  we  speak  of  quantitative  differ- 
ences— size,  yield,  or  of  rate  difTerences — we  are  really  concerned  with  differ- 
ences in  the  level  of  metabolism.  We  must  recognize  that  these  differences  in 
the  level  of  metabolism  are  bound  to  vary  against  different  environmental 
backgrounds,  and  where  the  particular  genes  involved  are  associated  with 
different  genetic  backgrounds. 

Our  approach  to  the  heterosis  problem  has  been  complicated  by  common 


PHYSIOLOGY  OF  GENE  ACTION  IN  HYBRIDS  113 

insistence  upon  attempts  to  find  a  single  genetic  mechanism.  It  has  suffered, 
too,  from  faikire  to  recognize  that  between  the  gene  and  the  final  mature 
organism  there  lies  a  system  of  developmental  processes  of  great  complexity. 
The  complexity  of  this  system  is  formidable  but  it  surely  can  be  analyzed, 
at  least  with  respect  to  its  most  significant  features,  if  it  is  taken  part 
by  part. 

SUMMARY 

The  evidence  relating  to  heterosis  suggests  that  the  phenomenon  is  to  be 
explained  genetically  in  terms  of  various  recombination  effects.  In  some  cases, 
dominance  is  the  important  consideration,  while  in  other  cases,  hetero- 
zygosity must  be  considered.  In  any  event,  it  is  the  resulting  specific  gene 
action  which  lies  at  the  basis  of  the  physiological  advantage  or  advantages 
which  give  rise  to  hybrid  vigor.  One  or  many  genes  may  be  involved.  Con- 
siderations of  genetic  balance  and  genotype-environment  balance  are  im- 
portant. Probably  most  cases  of  heterosis  are  to  be  explained  physiologically 
in  terms  of  differences  in  the  more  fundamental  aspects  of  the  metabolic  pat- 
tern, particularly  those  concerned  with  enzyme,  auxin,  and  other  growth 
substance  activity  in  plants  and  with  enzyme  and  hormonal  activities  in 
animals. 

To  clarify  the  mechanism  further,  studies  must  be  concerned  primarily 
with  the  genetics  and  physiology  of  early  development.  We  have  been  con- 
cerned with  mature  characteristics  of  size  and  yield,  with  the  inheritance  of 
so-called  quantitative  genes,  and  with  analyses  by  the  classic  methods  of 
genetics.  These  studies  have  brought  us  close  enough  to  an  understanding 
of  the  phenomenon  of  heterosis  to  indicate  that  its  further  analysis  by 
techniques  now  at  hand  will  uncover  facts  of  tremendous  importance  for 
genetics,  physiology,  and  other  studies  of  development,  some  of  them  con- 
siderably afield  from  heterosis  itself. 


WILLIAM  J.  ROBBINS 

Columbia  University  and  New  York  Bofanical  Garden 


Chapter  7 

Hybrid  Nutrifionol 
Requirements 


Hybrid  vigor  has  been  recognized  for  more  than  a  century.  It  has  been  con- 
sidered from  a  genetic,  morphological,  developmental,  physiological,  and 
commercial  standpoint.  Although  a  great  deal  of  information  has  been  ac- 
cumulated about  the  phenomenon,  we  are  still  unable  to  define  exactly  why 
a  hybrid  grows  better  than  the  parents  from  which  it  comes. 

It  is  obvious  that  the  cause  is  physiological — the  hybrid  functions  more 
effectively  or  for  a  longer  period  of  time,  and  accumulates  a  greater  mass  of 
cell  substance.  Its  metabolic  efficiency  is  greater  (East,  1936).  It  would  be 
illuminating  if  we  could  locate  specifically  the  physiological  processes  which 
are  responsible  for  the  greater  vigor  of  the  hybrid — recognizing  that  they  may 
be  numerous  and  complex  rather  than  single  and  simple,  and  that  they  may 
not  be  the  same  for  all  examples  of  hybrid  vigor. 

For  many  years  I  have  been  interested  in  the  factors  which  determine  why 
one  plant  species,  variety,  or  strain  grows  slowly  in  a  given  environment 
where  another  flourishes.  I  have  dealt  mainly  with  microorganisms,  especial- 
ly the  filamentous  fungi,  because  the  external  env  ironment  can  be  more  easily 
controlled  and  photosynthesis  is  not  a  complicating  factor.  From  my  ex- 
perience, as  well  as  from  the  work  of  others,  it  is  clear  that  in  many  instances 
growth — the  accumulation  of  cell  substance — is  limited  by  the  efficiency  of 
the  organism's  metabolic  machinery,  especially  the  activity  of  one  or  more 
enzyme  systems.  Whether  this  concept  can  be  applied  also  to  the  phenome- 
non of  hybrid  vigor  is  still  to  be  determined.  However,  it  is  a  hypothesis 
which  deserves  exploration. 

Let  us  begin  with  a  simple  example  of  growth-limitation.  Aspergillus  niger 
grows  well  in  a  liquid  medium  of  sugar,  mineral  salts,  and  asparagine.  In  the 
same  medium  Phycomyces  Blakesleeanus  will  not  grow  at  all. 

114 


HYBRID  NUTRITIONAL  REQUIREMENTS  115 

Does  Phycomyces  fail  to  grow  in  the  basal  solution  because  of  the  absence 
of  something  essential  which  it  needs  for  growth,  or  because  of  the  presence 
of  something  detrimental?  Does  Aspergillus  niger  grow  in  the  basal  solution 
because  it  does  not  need  to  be  furnished  with  the  "essential"  substance,  or 
because  it  is  more  resistant  to  the  supposed  injurious  ingredient? 

For  the  example  cited,  we  have  a  definite  and  well  demonstrated  explana- 
tion. Phycomyces  fails  to  grow  in  the  basal  medium  because  it  requires  the 
vitamin,  thiamine — which  it  is  unable  to  make  from  sugar,  mineral  salts,  and 
asparagine.  Aspergillus  niger  also  needs  thiamine,  but  it  constructs  the  vita- 
min from  the  elementary  materials  present  in  the  basal  solution.  In  this  in- 
stance, therefore,  the  failure  to  grow  is  due  to  the  lack  of  something  es- 
sential for  growth;  namely,  thiamine,  the  precursor  of  co-carboxylase. 

This  is  not  an  isolated  example.  Many  species  of  fungi  grow  slowly,  or  not 
at  all,  in  a  basal  medium  because  of  their  inability  to  make  one  or  more  of  the 
essential  metabolites.  These  metabolites  may  include  various  vitamins, 
purine  and  pyrimidine  bases,  amino  acids,  fatty  acids,  or  substances  as  yet 
unidentified. 

ESSENTIAL  METABOLITES-RELATION  TO  GROWTH 

It  may  be  assumed  that  the  complex  chemical  compounds  which  make 
up  the  cell  substance  of  a  living  organism  are  constructed  by  the  organism 
from  simpler  compounds.  A  series  of  intermediate  chemical  compounds  are 
formed  between  the  original  simple  foods  and  nutrients  and  the  final  product, 
cell  substance.  This  step-wise  progression  from  simple  to  complex  is  made 
possible  by  a  series  of  enzymes,  also  made  by  the  organism,  which  operate  on 
each  stage  as  that  stage  is  completed.  Although  synthesis  is  likely  to  be 
emphasized  in  considering  growth,  there  are  other  subsidiary  processes — 
necessary  concomitants  for  the  building  up  of  new  cell  substance.  The  cata- 
bolic  processes  of  digestion  and  respiration  also  occur  in  steps,  and  are  made 
possible  by  the  action  of  a  series  of  enzyme  systems. 

Any  substance  playing  a  necessary  part  directly  or  indirectly  in  the  chain 
of  reactions  which  end  in  the  synthesis  of  new  cell  substance  is  an  essential 
metabolite.  Unless  each  essential  metabolite,  each  chemical  substance  in  the 
step-wise  process  of  growth,  each  enzyme  which  facilitates  the  chemical  re- 
actions concerned,  is  made  within  the  organism  or  supplied  from  without,  the 
series  is  interrupted.  New  cell  substance  is  not  made,  and  growth  does  not 
occur.  If  not  enough  of  an  essential  metabolite  is  made,  growth  will  be 
slowed. 

Of  course,  this  is  an  oversimplified  statement  of  a  very  complicated 
process.  The  reactions  concerned  in  growth  probably  do  not  occur  in  a 
straight  line.  Some  steps  may  be  bypassed  and  side  reactions  may  occur,  all 
of  which  may  affect  the  speed  and  character  of  the  growth  which  results. 

It  would  be  difficult  to  estimate  the  number  of  essential  metabolites  in- 


116 


WILLIAM  J.  ROBBINS 


volved  in  the  growth  of  even  the  simplest  organism,  or  to  put  a  limit  on  the 
number  for  which  some  organism  may  not  eventually  be  found  to  exhibit  a 
deficiency. 

Some  species  or  strains  exhibit  a  complete  deficiency  for  one  or  more 
essential  metabolites.  They  are  unable  to  synthesize  any  of  the  substances  in 
question  and  do  not  grow  unless  the  substances  are  supplied  in  the  medium 
in  which  they  are  cultivated  (Robbins  and  Ma,  1942).  Others  suffer  from 
partial  deficiencies,  that  is,  they  grow  slowly  in  the  absence  of  a  particular 


Fig.  7.1 — Growth  affected  by  complete  and  partial  deficiencies  for  essential  metabolites. 
Fungi  grown  on  mineral-dextrose  medium  containing  asparagine  and  purified  agar  and 
supplemented  as  follows:  (1)  no  addition;  (2)  thiamine;  (3)  pyridoxine;  (4)  biotin;  (5)  thia- 
mine and  pyridoxine;  (6)  thiamine  and  biotin;  (7)  pyridoxine  and  biotin;  (8)  all  three  vita- 
mins. .\bove,  Ceratosiomella  midtiannulata,  complete  deficiency  for  pyridoxine,  partial  for 
thiamine;  below,  C.  microspora,  complete  deficiency  for  thiamine,  biotin,  and  pyridoxine. 


essential  metabolite  but  more  rapidly  if  it  is  added  to  the  medium  (Fig.  7.1). 

For  example,  the  clone  of  excised  tomato  roots,  with  which  we  have 
worked  for  many  years,  suffers  from  a  complete  deficiency  of  thiamine  and  a 
partial  deficiency  of  pyridoxine.  It  will  not  grow  unless  the  medium  contains 
thiamine  or  its  equivalent.  When  pyridoxine  is  added  to  a  medium  contain- 
ing thiamine,  the  growth  of  the  excised  roots  is  markedly  increased. 

In  a  sugar,  mineral-salt  solution,  the  growth  of  our  clone  of  excised  tomato 
ri)ots  is  limited  by  its  ability  to  synthesize  thiamine.  In  a  thiamine  solution, 
growth  is  limited  by  the  ability  of  the  roots  to  synthesize  pyridoxine  (Robbins, 
1946).  We  have  not  been  able  to  define  what  limits  the  growth  of  the  root 
in  a  solution  which  contains  both  thiamine  and  pyridoxine.  Other  examples 


HYBRID   NUTRITIONAL   REQUIREMENTS  117 

of  partial  deficiencies  could  be  cited.  Their  effect  is  to  decrease  the  rate  of 
growth  but  not  to  inhibit  it  entirely. 

As  a  result  of  investigations  which  have  extended  over  the  past  decade  or 
two,  we  know  of  many  examples  in  which  poor  growth  or  failure  to  grow  in  a 
specific  environment  is  due  to  the  inability  of  the  organism  to  synthesize 
adequate  quantities  of  one  or  more  essential  metabolites.  The  metabolic 
machinery  lacks  a  part,  or  some  part  works  slowly,  with  the  result  that  the 
organism  does  not  make  sufficient  quantities  of  one  or  more  growth  essen- 
tials, and  unless  supplied  with  the  missing  materials  from  without,  grows 
slowly,  or  not  at  all. 

Not  all  instances  of  failure  to  grow  or  of  poor  growth  in  a  given  environ- 
ment are  explainable  on  the  basis  of  deficiencies  of  essential  metabolites.  In 
some  instances  growth  may  be  limited  by  autogenic  growth  inhibitors. 

AUTOGENIC  INHIBITORS 

Zalokar  (1948),  Emerson  (1947, 1948),  and  others  have  described  a  mutant 
strain  of  Neurospora  which  grows  poorly  at  high  temperatures.  Growth  oc- 
curs if  sulfonamide  is  added  to  the  medium.  One  might  conclude  that 
sulfonamide  acts  for  this  organism  as  an  essential  metabolite.  It  appears, 
however,  that  this  mutant  produces  growth  inhibitors  which  are  antagonized 
in  some  way  by  the  sulfonamide.  This  seems  to  be  an  example  of  poor  growth 
caused  by  the  accumulation  of  autogenic  growth  inhibitors,  and  not  because 
of  the  lack  of  an  essential  metabolite. 

Information  on  the  role  of  autogenic  inhibitors  in  limiting  growth  is  less 
specific  and  more  difficult  to  obtain  than  evidence  for  the  limitation  of  growth 
due  to  a  deficiency  of  an  essential  metabolite.  How  commonly  do  internally 
produced  inhibitors  reduce  growth?  What  is  the  nature  of  these  substances? 

From  the  investigation  of  antibiotic  substances  we  know  that  many  organ- 
isms form  metabolic  products,  highly  inhibitory  for  organisms  other  than 
themselves.  Do  they  also  produce  substances  which  limit  their  own  growth? 
The  role  of  autogenic  inhibitors  in  limiting  growth  deserves  much  more 
attention  than  it  has  received. 

It  is  well  known  that  minute  amounts  of  specific  chemical  compounds 
materially  modify  the  amount  and  nature  of  growth  in  plants.  Zimmerman 
and  Hitchcock  (1949)  treated  Kalanchoe  plants  with  small  amounts  of  the 
ortho,  para,  and  meta  forms  of  chlorophenoxyacetic  acid.  The  para  form 
caused  the  apical  meristem  to  develop  into  a  spathe-like  organ  which  could 
be  cut  off  and  rooted.  It  had  little  resemblance  to  Kalanchoe.  The  ortho  and 
meta  forms  of  this  compound  did  not  have  this  effect.  This  modification  was 
not  a  mutation.  The  effect  wore  off  as  the  chemical  in  the  plant  disappeared, 
and  the  Kalanchoe  eventually  returned  to  its  normal  growth  pattern.  If  the 
change  had  been  permanent,  we  would  have  been  inclined  to  call  it  a  muta- 
tion and  look  for  a  genie  explanation ;  i.e.,  look  for  a  gene  which  controlled  the 


118  WILLIAM  J.  ROBBINS 

production  of  para-chlorophenoxyacetic  acid.  We  might  say  that  this  com- 
pound and  the  Kalanchoe  plant  acted  temporarily  as  linked  genes. 

Many  other  kinds  of  abnormal  growth  in  plants  are  probably  the  result  of 
the  effect  of  minute  amounts  of  specific  chemical  compounds.  Insect  galls 
are  characterized  by  an  abnormal  but  specific  growth  pattern  superimposed 
on  normal  tissue  by  the  presence  of  a  foreign  living  organism.  It  seems  very 
likely  from  the  observations  of  Boysen  Jensen  that  the  abnormal  growth  of 
insect  galls  is  caused  by  specific  chemical  compounds  produced  by  the  larvae 
which  inhabit  the  galls. 

It  must  be  emphasized  that  growth  is  an  extremely  complex  process,  not 
just  a  series  of  chemical  reactions.  To  consider  it  as  such  is  admittedly  an 
oversimplification  giving  no  thought  to  the  organization  in  which  these  re- 
actions occur,  or  to  the  structural  elements,  physical  processes,  and  chemical 
reactions  which  must  play  a  role. 

The  concept  of  growth  as  a  series  of  catalyzed  reactions  is  useful  and 
stimulating,  however,  in  considering  the  role  of  essential  metabolites — 
especially  enzymes — and  the  action  of  inhibitors  and  minute  amounts  of 
specific  chemical  compounds. 

HYBRID  VIGOR 

Some  years  ago  I  attempted  to  determine  whether  hybrid  corn  contains  a 
greater  quantity  of  substances  which  stimulate  the  early  growth  of  Phyco- 
myces  Blakesleeanus  than  the  inbred  parents.  The  effect  of  extracts  of  air 
dry  grains  and  of  partially  germinated  grains  of  the  hybrid  corn  and  its  in- 
bred parents  was  determined  on  the  growth  of  Phycomyces  in  the  presence  of 
thiamine  (Robbins,  1940,  1941a). 

When  compared  on  the  basis  of  extract  per  grain,  I  found  that  the  extracts 
of  the  grains  of  the  hybrid  corn  gave  a  greater  dry  weight  of  mycelium  of 
Phycomyces  than  those  of  either  of  the  inbred  parents  (Fig.  7.2).  The  stimu- 
lating material  seemed  to  be  present  in  both  the  embryo  and  the  endosperm. 
Since  the  solution  in  which  the  beneficial  effects  of  the  extracts  were  exhibited 
contained  sugar,  asparagine,  mineral  salts,  and  thiamine,  it  appeared  that 
the  effect  was  produced  by  unidentified  growth  substances.  These  were 
termed  for  convenience,  factor  Z. 

After  estimating  the  amount  of  factor  Z  present — from  the  effects  of  the 
extracts  of  the  corn  grains  on  the  early  growth  of  Phycomyces  in  the  presence 
of  thiamine — the  following  generalities  seemed  permissible.  The  amount 
of  factor  Z  increased  with  the  time  of  the  germination  of  the  corn  grains,  at 
least  up  to  seventy-two  hours'  germination.  The  quantity  of  Z  was  greater 
per  endosperm  than  per  embryo,  and  was  greater  in  the  grains  of  the  hybrid 
than  in  those  of  either  parent.  The  amount  of  thiamine  and  its  intermediates 
in  the  embryo  and  endosperm  of  the  grains  of  the  hybrid  and  its  parents 
was  not  correlated  with  the  amount  of  factor  Z,  nor  did  the  amount  of  biotin 
in  the  extracts  appear  to  be  correlated  with  the  amount  of  factor  Z. 


HYBRID  NUTRITIONAL  REQUIREMENTS 


119 


These  results  suggest  that  there  is  present  in  the  grains  of  corn,  material 
which  stimulates  the  early  growth  of  Phycomyces  in  the  presence  of  thiamine, 
and  that  there  is  more  of  this  material  per  grain  in  heterotic  hybrids  than  in 
those  of  the  inbred  parents. 

Interpretation  of  these  results  depends  in  part  on  the  identity  of  factor  Z. 

100 


90 


80 


O 

D 
_1 

u 

O 

>- 

H 

>- 
tr 
a 


70 


60 


50 


40 


30 


20 


10 


025 


0.5 
ML.  CORN   GRAIN   EXTRACT 


Fig.  7.2 — Increase  in  dry  weight  of  Phycomyces  produced  by  extracts  of  air  dry  grains  of 

maize.  Extracts  added  to  medium  of  sugar,  minerals,  asparagine,  and  thiamine.  A  =  line 

4-8;  B  =  line  187;  C  =  985,  4-8  X  187;  D  =  995,  187  X  4-8.  1  ml.  extract  =  1  grain. 


Unfortunately,  we  do  not  know  what  factor  Z  is.  We  succeeded  in  dividing  it. 
We  demonstrated  that  factor  Z  is  multiple,  and  separated  it  into  a  fraction 
adsorbed  on  charcoal,  factor  Zi,  and  a  filtrate  fraction,  factor  Z2.  Factor  Zi 
was  identified  as  hypoxanthine.  Factor  Z2  may  be  a  mixture  of  amino  acids. 
Although  this  problem  is  left  in  an  uncertain  and  unsatisfactory  condi- 
tion, it  suggests  a  line  of  attack.  This  would  be  an  investigation  of  heterosis 
by  studying  the  efifect  of  extracts  of  parents  and  of  heterotic  hybrids  on  the 
growth  of  other  organisms.  This  may  serve  as  a  means  of  bioassay  for  favor- 
able or  unfavorable  growth  factors. 


120 


WILLIAM  J.  ROBBINS 


Vigor  in   Heterocaryons 

Observations  of  Dodge  (1942)  on  heterocaryosis  in  Neurospora  are  of 
interest  to  the  general  problem  of  heterosis.  Dodge  inoculated  three  petri 
dishes,  one  with  his  Dwarf  16  strain  of  Xeurospora  telrasperma,  one  with  race 
C-8,  and  the  third  with  mixed  mycelium  or  conidia  of  both  the  dwarf  and  the 
C-8  races.  He  observed  that  the  mycelium  of  the  mixed  culture  grew  much 
more  rapidly  and  produced  more  abundant  conidia  than  the  mycelium  of 
either  the  dwarf  or  the  C-8  races  (Fig.  7.3). 


Fig.  7.3 — Heterocaryotic  vigor  in  Neurospora  telrasperma.  Growth  in  34  hours  at  room 

temperature  in  petri  dishes.  The  myceUum  of  the  two  heterocaryotic  races  {16  -\-  C  4  and 

16  +  C8)  has  nearly  covered  the  medium  in  the  dishes;  C4  and  C8  have  not  grown  halfway 

across  the  medium  and  Dwarf  16  has  made  no  visible  growth. 


When  two  races  of  Xeurospora  telrasperma  are  grown  together,  there  is  a 
migration  of  nuclei  through  the  openings  at  the  points  of  hyphal  anas- 
tomoses. The  races  need  not  be  of  opposite  sex.  After  nuclear  migration,  the 
cells  of  the  resulting  mycelium  are  heterocaryotic.  They  contain  two  kinds  of 
haploid  nuclei.  The  greater  vigor  of  the  mixed  culture  referred  to  above  ap- 
pears to  be  the  result  of  the  presence  in  a  common  cytoplasm  of  two  kinds  of 
nuclei. 

Heterocaryotic  vigor  does  not  always  accompany  heterocaryosis.  Dodge 
(1942)  observed  heterocaryotic  vigor  when  the  two  races,  Dwarf  16  and  C-4, 
were  grown  together.  But  heterocaryosis  for  races  C-4  and  C-8  did  not  result 
in  increased  vigor  in  the  mixed  culture.  Not  all  dwarf  races  act  as  race  16 
does.  Some  of  them  evidence  heterocaryotic  vigor  with  both  C-4  and  C-8, 


HYBRID   NUTRITIONAL   REQUIREMENTS  121 

others  with  C-4  but  not  with  C-8,  and  still  others  develop  none  with  either 
C-4  or  C-8. 

Dodge  has  suggested  that  the  heterocaryotic  hybrid  may  synthesize  a  full 
quantity  of  growth  substances  or  essential  metabolites.  Whereas  the  growth 
of  each  of  the  parents  is  limited  by  their  inability  to  synthesize  adequate 
quantities  of  one  or  more  essential  metabolites. 

Dwarf  16,  for  example,  may  be  able  to  make  adequate  quantities  of  essen- 
tial metabolites  1,  2,  3,  and  4,  but  unable  to  construct  enough  of  5,  6,  7, 
and  8.  On  the  other  hand,  race  C-4  may  be  unable  to  synthesize  enough  of 
1,  2,  3,  and  4,  but  be  capable  of  producing  an  adequate  supply  of  5,  6,  7, 
and  8.  When  nuclei  of  the  two  races  are  brought  together  in  a  common 
cytoplasm,  the  essential  metabolites  synthesized  by  one  of  the  nuclear  com- 
ponents supplement  those  synthesized  by  the  other  component.  The  hetero- 
caryotic mycelium  is  then  supplied  with  adequate  quantities  of  all  the 
essential  metabolites  necessary  for  rapid  growth. 

We  have  tried  to  test  this  hypothesis  by  supplementing  with  various 
substances  the  medium  on  which  race  16  and  other  dwarf  races  were  grown. 
If  it  were  possible  to  increase  materially  the  growth  rate  of  the  dwarf  race  by 
supplements  in  the  medium,  without  introducing  the  heterocaryotic  condi- 
tion, the  limiting  factors  for  dwarfness  could  be  identified  and  the  stimulus 
involved  in  the  heterocaryotic  condition  identified. 

A  basal  agar  medium  containing  mineral  salts,  dextrose,  asparagine,  neo- 
peptone,  and  thiamine  was  supplemented  by  a  mixture  of  purine  and  pyrim- 
idine  bases;  by  a  vitamin  mixture  containing  PAB,  calcium  pantothenate, 
inositol,  nicotinic  acid,  pyridoxine,  riboflavin,  thiamine,  guanine,  hypoxan- 
thine,  and  2-methyl-l,  4-naphthohydroquinone  diacetate;  by  malt  extract, 
casein  hydrolysate,  cow's  milk,  dried  yeast,  choline,  a-tocopherol,  hemin, 
oleic  acid,  ascorbic  acid  (filtered  sterile),  coconut  milk,  Taka-diastase 
(filtered  sterile),  water  extracts  of  the  mycelium  of  Neurospora,  liver  ex- 
tracts (both  filtered  sterile  and  heated),  adrenal  cortical  extract  (unheated), 
estrogenic  substance,  progesterone,  anterior  pituitary  extract,  posterior 
pituitary  extract,  whey,  or  potato  extract. 

None  of  the  substances  or  combinations  of  them  as  used  increased  the 
growth  rates  of  any  of  the  dwarf  races  to  an  extent  adequate  to  explain 
heterocaryotic  vigor.  Some  beneficial  effects,  usually  noted  only  in  older  cul- 
tures, were  obtained  from  cow's  milk  and  from  liver  extract.  These  efifects 
were  not  sufficiently  marked  to  suggest  that  either  supplement  supplied  the 
missing  factors. 

We  were  unsuccessful,  therefore,  in  defining  the  factors  limiting  the 
growth  of  the  dwarf  races  and  conversely  those  effective  in  inducing  more 
rapid  growth  in  the  heterocaryotic  mycelium. 

Our  failure  may  be  explained  in  various  ways.  We  may  not  have  included 
in  our  various  supplements  the  missing  essential  metabolites.  These  metabo- 


HYBRID  NUTRITIONAL  REQUIREMENTS  123 

lites  may  be  non-diffusible  or  very  labile  substances  such  as  enzyme  pro- 
teins, which  could  only  be  introduced  into  the  cell  through  inserting  a  nucleus 
and  its  genes.  The  original  hypothesis  may  be  in  error.  We  may  not  be 
dealing  with  limiting  quantities  of  essential  metabolites  but  with  inhibitors. 
We  might  assume  that  the  growth  of  one  or  both  of  the  parents  is  limited  by 
autogenic  inhibitors,  and  the  presence  of  both  kinds  of  nuclei  in  a  common 
cytoplasm  results  in  the  neutralization  in  some  fashion  of  the  inhibitors. 

Emerson  (1948)  has  succeeded  in  producing  heterocaryons  in  which  one 
kind  of  haploid  nucleus  neutralizes  the  effect  of  the  other.  The  augmented 
growth  of  the  heterocaryon,  as  compared  to  that  of  strains  which  are 
homozygous,  reminds  one,  says  Emerson,  of  instances  of  single  gene  heterosis 
in  maize  reported  by  Jones. 

The  importance  of  internal  factors  in  heterosis  is  suggested  by  the  results 
I  obtained  on  the  growth  of  the  excised  roots  of  a  heterotic  tomato  hybrid 
and  its  inbred  parents  (Robbins,  1941b).  The  hybrid  roots  and  the  roots  of 
the  two  inbred  parents  were  grown  in  liquid  culture  which  contained  mineral 
salts  and  cane  sugar.  This  basal  medium  was  supplemented  with  thiamine, 
with  thiamine  and  pyridoxine,  and  with  thiamine,  pyridoxine,  and  nicotina- 
mide. 

Growth  of  the  roots  of  the  hybrid  exceeded  that  of  either  of  the  inbred 
parents  in  all  three  types  of  media  (Fig.  7.4).  Growth  of  one  parent  was  im- 
proved by  the  addition  of  pyridoxine  to  the  thiamine  solution,  but  a  further 
supplement  of  the  medium  with  nicotinamide  had  little  effect.  Growth  of  the 
second  inbred  parent  was  little  affected  by  the  addition  of  pyridoxine  to  the 
thiamine  medium,  but  was  improved  by  the  further  addition  of  nicotinamide 
to  the  thiamine  and  pyridoxine  solution. 

These  results  suggest  that  the  greater  vigor  of  growth  of  the  heterotic 
hybrid  is  determined  in  part  by  its  greater  ability  to  synthesize  pyridoxine 
and  nicotinamide.  That  is  evidently  not  the  whole  story,  because  its  growth 
exceeded  that  of  the  inbred  parents  in  media  containing  all  three  vitamins. 

Although  heterosis  may  be  considered  and  should  be  considered  from  the 
genetical  standpoint,  it  should  also  be  studied  from  the  physiological  stand- 
point. I  have  suggested  that  it  may  be  important  to  devote  attention  to  the 
question  of  v/hat  the  internal  factors  are  which  limit  growth,  what  they  are 
in  inbreds,  and  how  they  are  removed  in  heterotic  hybrids.  We  should  con- 
sider in  such  investigations  the  role  of  essential  metabolites,  of  growth  in- 
hibitors, and  of  other  specific  chemical  compounds  which  materially  modify 
growth.  Microorganisms  might  be  utilized  as  tools  for  the  detection  of  growth 
stimulators  or  growth  inhibitors. 


EDGAR  ANDERSON 

Missouri  Bofanical  Garden 

and 

WILLIAM  L.  BROWN 

Pioneer  Hybrid  Corn  Company 


Chapter  8 

Origin  of  Corn  Belt  Moize  and  Its 

Genetic  Significance 


Several  ends  were  in  view  when  a  general  survey  of  the  races  and  varieties  of 
Zea  mays  was  initiated  somewhat  over  a  decade  ago  (Anderson  and  Cutler, 
1942).  Maize,  along  with  Drosophila,  had  been  one  of  the  chief  tools  of  mod- 
ern genetics.  If  one  were  to  use  the  results  of  maize  genetics  most  efficiently 
in  building  up  general  evolutionary  theories,  he  needed  to  understand  what 
was  general  and  what  was  peculiar  in  the  make-up  of  Zea  mays.  Secondly, 
since  maize  is  one  of  the  world's  oldest  and  most  important  crops,  it  seemed 
that  a  detailed  understanding  of  Zea  mays  throughout  its  entire  range  might 
be  useful  in  interpreting  the  histories  of  the  peoples  who  have  and  are  using 
it.  Finally,  since  maize  is  one  of  our  greatest  national  resources,  a  survey  of  its 
kinds  might  well  produce  results  of  economic  importance,  either  directly  or 
indirectly. 

Early  in  the  survey  it  became  apparent  that  one  of  the  most  significant 
sub-problems  was  the  origin  and  relationships  of  the  common  yellow  dent 
corns  of  the  United  States  Corn  Belt.  Nothing  exactly  like  them  was  known 
elsewhere  in  the  world.  Their  history,  though  embracing  scarcely  more  than 
a  century,  was  imperfectly  recorded  and  exasperatingly  scattered.  For  some 
time  it  seemed  as  if  we  might  be  able  to  treat  the  problem  only  inferentially, 
from  data  derived  from  the  inbred  descendants  of  these  same  golden  dent 
corns.  Finally,  however,  we  have  been  able  to  put  together  an  encouragingly 
complete  history  of  this  important  group  of  maize  varieties,  and  to  confirm 
our  historical  research  with  genetical  and  cytological  evidence. 

An  even  approximate  survey  of  Zea  way.s-as-a-whole  remains  a  goal  for 

124 


ORIGIN   AND  SIGNIFICANCE  OF  CORN   BELT  MAIZE  125 

the  distant  future,  but  our  understanding  of  Corn  Belt  dent  corns  is  already 
more  complete  than  we  had  originally  hoped.  Since  our  evidence  is  detailed 
and  of  various  kinds,  it  may  make  the  presentation  somewhat  easier  to  follow 
if  we  give  a  brief  description  of  the  pre-hybrid  commercial  yellow  dents  of 
the  United  States  Corn  Belt,  review  their  history  in  broad  outline,  and  then 
proceed  to  an  examination  of  the  various  kinds  of  evidence  on  which  these 
generalizations  have  been  built. 

Corn  Belt  dents,  the  commercial  varieties  which  dominated  the  chief 
centers  of  corn  production  in  the  United  States  for  over  half  a  century  pre- 
ceding the  advent  of  hybrid  corn,  were  variable  open-pollinated  varieties. 
They  varied  from  plant  to  plant,  from  field  to  field  of  the  same  variety,  and 
from  variety  to  variety.  Figure  8.1,  based  upon  an  examination  of  a  field  of 
Golden  Queen,  one  of  the  lesser  known  of  these  varieties,  will  indicate  the 
kind  of  variation  which  characterized  the  fields  of  that  day. 

In  spite  of  this  variation,  or  one  might  almost  say,  impressed  on  top  of  it, 
was  a  remarkably  persistent  combination  of  generally  prevalent  characters. 
Considered  from  plant  to  plant  or  from  field  to  field,  as  individuals,  these 
varieties  seemed  ephemeral  and  unimportant.  Seen  as  populations,  as  col- 
lections of  inter-breeding  individuals,  the  Corn  Belt  dents  as  a  whole  were  a 
well-marked  and  definite  entity,  particularly  when  contrasted  with  maize 
in  other  parts  of  the  world.  They  tended  to  have  one  well-developed  ear,  fre- 
quently accompanied  by  a  small  ear  at  the  node  below  this  primary  one. 
The  ears  had  large,  nearly  cylindrical  cobs  with  red  or  reddish  glumes.  The 
usually  golden  yellow  kernels,  pronouncedly  dented  at  the  tip,  had  a  peri- 
carp frequently  roughened  by  tiny  wrinkles.  They  were  set  in  from  14  to  22 
straight  rows  with  little  external  indication  of  the  fact  that  the  rows  were  in 
pairs.  The  mathematical  perfection  of  the  ear  was  frequently  lessened  by  a 
slight  tendency  for  the  whole  ear  to  taper  toward  the  apex,  and  for  the  row- 
ing of  the  kernels  and  the  diameter  of  the  cob  to  be  somewhat  differentiated 
in  its  lowermost  quarter. 

Characteristically,  the  plant  on  which  this  ear  was  borne  had  a  single,  up- 
right stem,  leaves  with  tight  sheaths  and  strong,  arching  blades,  and  a 
heavy,  many -branched  tassel.  Kernel  color  was  remarkably  standardized, 
a  faint  flush  of  coppery  red  in  the  pericarp  and  a  yellow  endosperm,  combin- 
ing to  give  varying  shades  of  deep,  golden  color.  Epidermal  color  was  ap- 
parent on  the  culm  and  leaves  at  the  base  of  the  plant,  but  seldom  or  never 
were  there  to  be  found  the  brilliant  reds,  dark  purples,  and  other  foliage 
colors  which  are  so  characteristic  of  maize  in  various  parts  of  Latin  America. 
While  there  was  some  variation  in  anther  color  and  silk  color,  pinks  and  dull 
reds  were  commonest  though  greens  and  bright  reds  were  not  unknown. 

As  we  have  shown  elsewhere  (Anderson  and  Brown,  1950)  there  cannot 
be  the  slightest  doubt  that  these  widespread  and  standardized  Corn  Belt 
varieties  were  the  creation  of  the  nineteenth  century.  They  came  in  large  part 


20 


18 


OS. 
Ul 


C9 


16 


% 


^ 


^•V  i.  *- 


14 


12 


8-9 

10- 

■11                12- 

13 

14-15 

KERNEL  WIDTH 

TASSEL  BRANCH 
NUMBER 

GLUME  LENGTH 
IN  MM. 

PITH  WIDTH 
IN  MM. 

EAR  LENGTH 
IN  CM. 

•  11-20 

•  8-9 

•4-9 

•26-30 

•  21-23 

4io 

4io-ii 

•  22-25 

#^OVER  24 

•  11-13 

•  12-15 

•  14-21 

Fig.  8.1 — Pictorialized  diagram  showing  relationship  between  numbers  of  rows  of  kernels, 

kernel  width,  tassel  branch  number,  glume  length,  pith  diameter,  and  ear  length  in  an 

open-pollinated  sample  of  Golden  Queen  dent  corn. 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE  127 

from  crosses  between  White  Southern  Dents,  mostly  of  Mexican  origin,  and 
the  long,  slender  Northern  Flints  which  had  dominated  the  eastern  United 
States  for  at  least  some  hundreds  of  years  preceding  the  discovery  of 
America.  While  these  two  complexes  were  of  primary  importance  in  the  crea- 
tion of  Corn  Belt  corn,  it  should  be  pointed  out  that  germ  plasm  of  other 
types  of  maize  has  undoubtedly  filtered  into  Corn  Belt  mixtures.  Compared 
to  Southern  Dents  and  Northern  Flints,  these  certainly  are  of  minor  im- 
portance. There  are,  nevertheless,  to  be  found  among  dent  inbreds  of  the 
Corn  Belt  certain  strains  which  exhibit  Caribbean  influence  and  others 
which  seem  to  contain  germ  plasm  of  southwestern  United  States  or  western 
Mexican  varieties. 

Although  the  following  discussion  does  not  go  into  detail  regarding  the 
influence  of  these  secondary  sources  of  germ  plasm  on  Corn  Belt  corn,  the 
effects  of  such  influences  are  important  and  we  have  already  made  small 
beginnings  at  studying  them.  The  Northern  Flints  are  in  some  ways  strik- 
ingly similar  to  the  common  yellow  flints  of  the  Guatemalan  highlands,  strik- 
ingly unlike  most  Mexican  maize.  They  are  one  of  several  cultural  traits 
which  apparently  spread  from  the  Mayan  area  to  the  eastern  United  States 
without  leaving  any  clear  record  of  the  route  by  which  they  came.  In  their 
general  appearance,  as  well  as  in  technical  botanical  details,  the  Northern 
Flints  were  very  different  from  the  Southern  Dents.  The  hybrid  vigor  which 
resulted  from  mixing  these  diverse  types  was  soon  noted  by  alert  agricultur- 
ists. While  some  of  the  blending  of  flints  and  dents  may  have  been  haphazard 
and  accidental,  much  of  it  was  directed  and  purposeful.  The  benefits  to  be 
gained  were  listed  in  public,  and  the  exact  effects  of  continued  mixing  and  of 
backcrossing  were  discussed  in  detail  as  early  as  1825  (Lorain,  1825).  This 
intelligent,  controlled  hybridizing  proceeded  for  at  least  a  half  century  until 
the  new  yellow  dents  were  so  ubiquitous  and  everyday  that  their  very  origin 
was  forgotten. 

For  theoretical  reasons  this  neglect  of  historical  tradition  was  unfortunate. 
Maize  breeders  have  not  understood  that  the  heterosis  they  now  capitalize 
is  largely  the  dispersed  heterosis  of  the  open-pollinated  flint-dent  mongrels. 
Maize  geneticists  are  for  the  most  part  unaware  that  the  germ  plasm  they 
use  for  fundamental  generalizations  is  grossly  atypical  of  germ  plasms  in 
general.  We  shall  return  to  a  detailed  discussion  of  these  two  points  after 
referring  briefly  to  the  evidence  concerning  the  origin  of  Corn  Belt  maize. 

Though  there  is  abundant  evidence  that  our  Corn  Belt  dents  came  from 
mixtures  of  Northern  Flints  and  Southern  White  Dents,  the  evidence  con- 
cerning these  two  regional  types  is  very  one-sided.  The  Northern  Flints 
(Brown  and  Anderson,  1947)  were  remarkably  uniform  from  place  to  place 
and  from  century  to  century.  The  archaeological  record  is  rich  going  back  to 
early  pre-Columbian  times  and  there  are  numerous  naive  but  accurate  de- 
scriptions of  these  varieties  in  colonial  accounts. 


128  EDGAR  ANDERSON  AND  WILLIAM  L.  BROWN 

The  Southern  Dents  (Brown  and  Anderson,  1948)  are  much  more  vari- 
able. For  over  a  century  their  variability  has  been  stressed  by  all  those  who 
have  discussed  them.  The  samples  which  we  obtained  from  the  South  differed 
from  field  to  field,  and  from  variety  to  variety.  For  an  accurate  understanding 
of  them  and  their  history,  we  would  like  many  more  archaeological  specimens 
than  we  have  for  the  flints,  and  many  more  colonial  descriptions.  Instead,  we 
have  as  yet  no  archaeological  record,  merely  two  accounts  in  early  colonial 
times — one  from  Louisiana  and  the  other  from  Virginia.  There  is  one  passing 
mention  in  a  pre-revolutionary  diary,  and  then  a  truly  remarkable  discussion 
by  Lorain  in  1825.  Finally,  the  United  States  Patent  Office  report  for  1850 
gives  us,  for  region  after  region,  a  detailed  picture  of  the  extent  to  which  this 
purposeful  mixing  had  proceeded  by  that  time. 

To  summarize  the  historical  evidence,  the  Northern  Flints  were  once  the 
prevailing  type  of  maize  throughout  the  eastern  United  States  (Brown  and 
Anderson,  1947)  with  an  archaeological  record  going  back  at  least  to  a.d. 
1000.  There  is  as  yet  no  archaeological  evidence  for  their  having  been  pre- 
ceded in  most  of  that  area  by  any  other  type  of  maize,  or  of  Mexican-like 
dents  having  been  used  there  in  pre-Columbian  times.  The  Northern  Flints 
belong  to  a  type  of  maize  rare  or  unknown  over  most  of  Mexico,  but  common 
in  the  highlands  of  Guatemala.  The  Southern  Dents,  on  the  contrary,  obvi- 
ously are  largely  derived  from  Mexican  sources,  and  by  1700  were  being 
grown  as  far  north  as  Louisiana  and  Virginia  (Brown  and  Anderson,  1948). 
As  to  how  and  when  they  spread  north  from  Mexico,  we  have  no  evidence 
other  than  the  negative  fact  that  they  are  not  known  archaeologically  from 
the  eastern  United  States,  and  are  not  represented  in  the  collections  of  early 
Indian  varieties  from  that  region. 

As  early  as  1800,  the  benefits  of  crossbreeding  these  two  different  types  of 
maize  were  appreciated  by  at  least  a  few  experts.  By  1850  the  process  was 
actively  under  way  from  Pennsylvania  to  Iowa,  and  south  to  the  Gulf  states. 
By  the  '70's  and  '80's,  a  new  type  of  corn  had  emerged  from  this  blending, 
although  crossing  and  re-crossing  of  various  strains  continued  up  to  the  ad- 
vent of  hybrid  corn.  During  the  latter  half  of  the  process,  the  origin  of  Corn 
Belt  dents  from  50  to  100  generations  of  selective  breeding  of  crosses  of 
Northern  Flints  and  Southern  Dents  was  almost  completely  forgotten.  Hav- 
ing at  length  resurrected  the  evidence  (Anderson  and  Brown,  1950)  for  this 
mingling  of  two  fundamentally  different  types  of  maize,  we  shall  now  turn 
to  the  genetical  and  cytological  evidence  which  first  called  the  phenomenon 
to  our  attention  and  led  us  to  search  for  historical  proof. 

CYTOLOGY 

The  most  important  cytological  contribution  on  the  origin  of  Corn  Belt 
maize  is  found  in  a  comparison  of  the  numbers  and  distribution  of  chromo- 
some knobs  in  the  Northeastern  Flints,  open-pollinated  varieties  of  Southern 


ORIGIN   AND  SIGNIFICANCE  OF  CORN   BELT  MAIZE  129 

Dents,  and  inbred  strains  of  Corn  Belt  dents.  As  has  been  shown  previously 
(Longley,  1938)  and  (Reeves,  1944),  chromosome  knobs  may  be  an  im- 
portant tool  in  studying  relationshii)S  in  maize.  Our  work  with  North  Ameri- 
can corn  not  only  supports  this  contention,  but  suggests  that  knob  data  may 
be  even  more  important  than  has  previously  been  supposed. 

The  8-10  rowed  flint  and  flour  varieties  of  New  York,  Pennsylvania,  and 
New  England  are  nearly  knobless.  In  the  material  we  have  examined,  they 
have  0  to  2  knobs.  These  observations  are  in  agreement  with  Longley's 
earlier  conclusions  that  maize  varieties  of  the  northern  Indians  were  char- 
acterized by  having  few  knobs.  Longley's  material,  however,  included  no 
strains  from  northeastern  United  States — the  area  in  which  the  flint  an- 
cestors of  Corn  Belt  corn  were  highly  concentrated.  It  is  interesting,  more- 
over, to  note  that  varieties  from  this  segment  of  North  America  have  even 
fewer  knobs  than  do  the  strains  from  most  Northern  Plains  Indian  tribes. 

In  contrast,  many  more  knobs  were  to  be  found  in  the  open  pollinated 
varieties  of  Southern  Dent  corn.  In  these  strains  we  have  found  numbers 
ranging  from  5  to  12,  for  those  varieties  representing  the  least  contaminated 
segment  of  present-day  Southern  Dent  corn.  These  cytological  data  are  in 
complete  agreement  with  the  known  facts  regarding  the  history  of  Northern 
Flints  and  Southern  Dents. 

There  seems  little  doubt  that  the  Gourdseed-like  Dents'  of  the  southeast- 
ern United  States  have  stemmed  directly  from  Mexico  wiiere  morphological- 
ly and  cytologically  similar  corns  can  be  found  even  today.  Likewise,  we 
have  found  in  highland  Guatemala  varieties  of  maize  with  ear  character- 
istics strikingly  similar  to  Northern  Flints  and  with  as  few  as  three  knobs. 
Insofar  as  cytology  is  concerned,  therefore,  it  is  not  at  all  difficult  to  visualize 
a  Guatemalan  origin  for  Northeastern  Flint  corn.  The  Corn  Belt  inbreds 
with  which  we  have  worked  (Brown,  1949)  have  knob  numbers  of  1  to  8. 
The  distribution  of  numbers  in  these  strains  is  almost  exactly  intermediate 
between  that  of  Northern  Flints  and  Southern  Dents  (Fig.  8.2).  This  evi- 
dence, based  on  a  character  which  certainly  has  not  been  intentionally 
altered  by  selection,  strongly  fortifies  the  archaeological  and  historical  facts 
pointing  to  a  hybrid  origin  of  Corn  Belt  dent  corns. 

GENETIC  EVIDENCE 

The  genetical  evidence  for  the  origin  of  Corn  Belt  maize  from  mixtures 
of  Northern  Flints  and  Southern  Dents  is  of  various  kinds.  In  its  totality,  it 
is  so  strong  that,  had  we  not  been  able  to  find  the  actual  historical  evidence, 
we  could  have  determined  what  had  happened  from  genetic  data  alone.  In 
the  first  place  we  have  demonstrated,  by  repeating  the  cross,  that  it  is  pos- 
sible to  synthesize  Corn  Belt  dents  from  hybrids  between  Southern  Dents 

I  The  name  "Gourdseed"  has  been  used  since  colonial  times  to  describe  the  extremely- 
long  seeded,  white  Southern  Dents,  whose  kernels  are  indeed  not  so  diflferent  in  appear- 
ance from  the  seeds  of  gourds  of  the  genus  Lagenaria. 


130 


EDGAR  ANDERSON  AND  WILLIAM  L.  BROWN 


and  Northern  Flints.  Our  experiments  in  crossing  a  typical  white  gourdseed 
from  Texas  and  a  typical  yellow  flint  from  New  York  State  are  now  only  in 
the  third  generation  and  are  being  continued.  However,  it  is  already  evident 
that  some  of  the  segregates  from  this  cross  are  within  the  range  of  varia- 
tion of  Corn  Belt  dents  (Fig.  8.3). 

In  spite  of  the  50  to  100  generations  of  mixing  which  has  taken  place,  the 
characters  of  Northern  Flints  and  Southern  Dents  still  tend  to  be  associated 
in  Corn  Belt  dents.  Anderson  (1939)  has  shown  that  in  crosses  between  species 


10 


OLD     SOUTHERN     OENTS 


10 

> 
u 

Z     0 
lit 

3 

a 

^20 
li. 


10 


DERIVED    SOUTHERN    OENTS 


CORN     BELT     INBREDS 


10 


1 

NOR 

1 

THERN    ruiNTS 

0 

1     '     2 

'    3 

4     ' 

5    ' 

6 

7 

8    •■ 

9    ' 

10  ' 

II 

'    12 

"I 

NUMBER  OF  CHROMOSOME   KNOBS 

Fig.  8.2 — Frequency  distribution  of  chromosome  knobs  in  Northern  Flints,  Southern 

Dents,  and  Corn  Belt  inbreds. 


or  between  races,  all  the  multiple  factor  characters  which  characterize  each 
are  partially  linked  with  one  another  and  tend  to  remain  associated,  even 
after  generations  of  controlled  breeding.  More  recently  he  has  used  this 
principle  in  the  development  of  the  method  of  extrapolated  correlates  (Ander- 
son, 1949)  by  which  the  original  characteristics  can  be  deduced  from  the  mix- 
tures even  when  previously  unknown. 

Using  this  method  in  a  relatively  crude  form,  we  were  able  (in  advance 
of  our  historical  evidence)  to  demonstrate  (Brown,  1949)  in  Corn  Belt  in- 
breds, the  association  of  low  knob  numbers,  flag  leaves,  cylindrical  ears,  few 
tassel  branches,  and  flinty  kernels — all  characteristics  which  typify  the 
Northern  Flints.  Similarly,  it  was  possible  to  show  the  association  among 
these  98  Corn  Belt  inbreds  of  high  knob  numbers,  no  flag  leaves,  tapering 
ears,  dented  kernels,  and  many  tassel  branches — a  combination  of  char- 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE  131 

acters  which  is  typical  of  the  Southern  Dents.  As  a  matter  of  fact,  by  this 
technique  Brown  predicted  the  knob  numbers  of  the  Northern  Flints,  even 
when  that  fact  was  unknown  to  us. 

The  association  of  characters  in  actual  open-pollinated  fields  of  Corn  Belt 
dents  is  so  complex  that  one  might  suppose  any  study  of  it  would  be  hopeless. 
However,  from  a  study  of  character  association  in  an  open-pollinated  field 


Fig.  8.3 — Corn  Belt  Denl-like  segregates  from  an  Fo  generation  of  cross  of  Longfellow 

Flint  X  Gourdseed  Dent. 

of  Golden  Queen  Dent  (Fig.  8.1)  we  were  able  to  demonstrate  the  association 
of:  (1)  wide  kernels,  (2)  low  row  numbers,  (3)  short  glumes,  (4)  few  tassel 
branches,  (5)  long  ears,  and  (6)  narrow  central  pith  in  the  ear — all  of  these 
characterizing  Northern  Flints.  The  opposing  combination:  (1)  narrow 
kernels,  (2)  high  row  numbers,  (3)  long  glumes,  (4)  many  tassel  branches, 
(5)  short  ears,  and  (6)  wide  central  pith  also  tended  to  be  associated  and  is 
characteristic  of  Southern  Dents.  In  other  words,  some  of  the  characters 
which  went  in  together  from  flints  and  dents  were  still  in  this  open-pollinated 
variety  tending  to  stay  together  on  the  average.  The  existence  of  such  char- 
acter complexes  has  been  appreciated  by  experienced  corn  breeders,  though 
apparently  it  has  never  been  commented  on  in  print.  Of  course,  corn  breed- 
ers and  corn  geneticists  differ  in  their  endowments  for  apprehending  such 


132  EDGAR  ANDERSON  AND  WILLIAM  L.   BROWN 

phenomena  in  advance  of  the  published  facts,  and  the  existence  of  these 
strong  linkages  has  been  more  apparent  to  some  than  to  others. 

WIDTH  OF  CROSS 

The  demonstration  that  Corn  Belt  dents  largely  are  derived  from  hy- 
bridization between  Southern  Dents  and  Northern  Flints  is  of  particular  im- 
portance because  this  is  such  a  wide  cross.  Our  evidence  for  this  assertion  is 
largely  morphological,  though  there  is  supporting  evidence  from  cytology 
and  genetics. 

In  nearly  all  species  of  cultivated  plants  there  are  conspicuous  differences 
in  color  and  shape.  These  differences  give  the  various  cultivated  varieties  of 
a  species  a  false  aspect  of  difference  from  one  another,  and  from  their  wild 
progenitors.  False,  because  these  differences  are  usually  due  to  a  few  genes,  if 
not  being  actually  monofactorial.  The  striking  differences  between  such 
varieties  are  therefore  no  true  indication  of  the  distinctness  of  their  germ 
plasms. 

On  the  other  hand,  there  are  subtle  differences  in  form,  proportion,  and 
indument  which,  though  difficult  for  a  novice  to  apprehend,  are  more  like 
the  differences  which  distinguish  distinct  species  of  the  same  genus.  These 
taxonomically  important  differences  have  proven  valid  criteria  for  indicating 
the  diversity  of  germ  plasms.  So  it  has  been  proven  that  the  subtle  taxonomic 
differences  between  the  Old  World  and  New  World  cottons  are  much  more 
representative  of  the  genetic  diversity  and  relationships  of  these  two  groups 
of  varieties  than  are  the  conspicuous  differences  in  color  and  leaf-shape  which 
are  found  within  each  group.  In  the  Cucurbits  the  striking  differences  in 
color  and  form  of  fruit,  which  differentiate  the  varieties  of  Cucurbita  Pepo 
and  of  C.  moschata,  are  superficial  compared  to  the  taxonomically  significant 
features  which  separate  these  two  groups.  The  latter,  moreover,  have  been 
proved  to  be  a  significant  index  of  genetic  diversity,  either  between  these 
two  groups  of  Cucurbits  or  in  assaying  the  variation  within  C.  Pepo  itself 
(Shifriss,  1947)  (Whitaker  and  Bohn,  1950). 

The  difficulty  in  relying  upon  such  taxonomic  criteria  is  that  the  method 
is  highly  subjective.  Taxonomy  is  of  necessity  still  more  of  an  art  than  a 
science.  This  means  that  one  must  personally  examine  the  evidence  if  his 
opinion  is  to  be  worth  anything.  It  also  means  that  the  worker's  opinion  is 
worth  no  more  than  his  understanding  of  the  taxonomic  entities  included  in 
his  judgment.  However,  until  more  objective  criteria  are  evolved  for  this 
field,  we  shall  have  to  use  fairly  traditional  taxonomic  methods  for  want  of 
anything  better.  Accordingly,  the  senior  author  has  for  two  years  spent  one 
day  a  week  in  a  technical,  agrostological,  herbarium  survey  of  all  the  grasses 
conceivably  related  to  Zea  mays — all  the  genera  in  the  tribes  Andropogoneae 
and  Maydeae.  With  that  background,  his  judgments  may  well  be  mistaken 
but  they  are  certainly  informed. 

From  this  point  of  view,  the  variation  within  Zea  mays  is  without  parallel, 


ORIGIN  AND  SIGNIFICANCE  OF  CORN   BELT  MAIZE  133 

not  only  in  the  cultivated  cereals  but  in  any  other  domesticated  plant  or 
animal.  There  are  such  superficial  characters  as  aleurone  color,  pericarp 
color,  plant  color,  carbohydrate  composition,  and  such  amazing  single  factor 
differences  as  tunicate  and  teopod.  In  addition,  there  are  a  whole  battery  of 
characters  which  are  difficult  to  work  with  genetically,  but  which  are  the 
kinds  of  differences  that  agrostologists  find  significant  in  the  deployment  of 
species  and  genera:  spikelet  shape  and  venation,  spikelet  arrangement, 
rachis  morphology,  pubescence,  leaf-shape,  internode  proportions,  etc.  Using 
such  criteria,  the  hybridization  of  the  Southern  Dents  and  the  Northern 
Flints  represents  the  mingling  of  two  basically  different  germ  plasms. 

For  evidences  of  relationship,  the  male  inflorescence  of  maize  (the  tassel) 
is  of  particular  importance.  Inflorescence  differences  generally  have  proved 
to  be  of  primary  taxonomic  importance  in  the  Gramineae.  Variation  in  the 
male  inflorescence  of  Zea  would  likely  be  less  obscured  by  domestication  than 
the  female  inflorescence  (the  ear)  which  has  been  deliberately  selected  for 
various  peculiarities.  The  entire  male  inflorescence  of  the  Southern  Dents 
has  been  extensively  modified  by  condensation  (Anderson,  1944),  a  sort  of 
fasciation  which  telescopes  adjacent  nodes,  and  in  the  ear  produces  increases 
in  row  number.  It  is  an  abnormality  conditioned  by  at  least  two  pairs  of 
recessive  genes  and  its  expression  is  certainly  modified  by  still  other  genes. 

Tassels  of  the  Northern  Flints  are  without  any  condensation.  Though 
condensation  modifies  the  general  aspect  of  the  tassel,  it  is  relatively  super- 
ficial. The  presence  of  so  much  condensation  renders  difficult  the  demonstra- 
tion of  a  much  more  fundamental  difference.  The  central  spike  of  the  North- 
ern Flints  is  decussately  arranged.  That  is,  the  pairs  of  spikelets  are  in  alter- 
nate whorls  of  two;  whereas  the  spike  of  the  Southern  Dents  (allowing  for 
the  modifications  produced  by  extreme  condensation)  is  fundamentally  in 
whorls  of  3,  or  mixtures  of  whorls  of  3  and  whorls  of  2.  The  rachis  of  the 
Northern  Flints  is  slender  with  long  internodes,  that  of  the  Southern  Dents 
is  short  and  flattened  (Fig.  8.5).  Pedicels  of  the  upper  spikelets  always  are 
long  in  the  Northern  Flints.  In  the  Southern  Dents  they  may  be  so  short  that 
one  cannot  distinguish  the  normally  pedicellate  spikelet  from  its  sessile 
partner. 

Correlated  differences  are  seen  in  the  ear.  That  of  the  Northern  Flints  has 
a  narrow  central  pith  and  is  long  and  slender,  characteristically  with  8-10 
rows.  The  ear  of  the  Southern  Dents  is  short  and  thick  with  a  wide  central 
pith,  and  with  from  16  to  30  or  more  rows.  Pairing  of  the  rows  is  markedly 
evident  in  the  Northern  Flints,  even  when  they  are  pushed  closer  together 
in  those  occasional  ears  with  10  or  12  rows  (Fig.  8.4).  There  is  little  or  no  row 
pairing  in  the  Southern  Dents.  The  kernel  of  the  Southern  Dents  is  long,  flat, 
and  narrow.  Its  largest  diameter  is  near  the  base.  By  contrast,  the  kernel  of 
the  Northern  Flints  is  wider  than  it  is  high,  and  is  considerably  thicker 
at  the  apex  than  it  is  at  the  base. 

The  ear  of  Zea  mays  is  terminal  on  a  secondary  branch,  which  is  hidden  by 


.^     r^ 


-n 


-.»,  -iii*;  iafe. 


S-2 


M. 


S-l 


N-3 


N-2 


N-4 


t-  '■ 


S-4- 


*•-. 


S-3 


Fig.  8.4 — Typical  ears  (/),  shanks  (2),  and  seeds  {3  and  4)  of  Northern  Flint  (iV),  and 

Southern  Dent  (5). 


Fig.  8.5 — Typical  plants,  tassels,  and  staminate  spikelets  of  Northern  Flint  and  Southern 

Dent. 


136  EDGAR  ANDERSON  AND  WILLIAM  L  BROWN 

its  specialized  leaves  or  husks.  When  dissected  out,  these  ear  shoots  (or 
shanks)  are  diagnostically  different  in  Northern  Flints  and  Southern  Dents 
(Fig.  8.4).  In  the  former  they  are  long,  with  elongated  internodes  which  are 
widest  between  the  nodes,  and  which  have  a  smooth  surface  upon  drying.  In 
the  latter  they  are  very  short,  frequently  wider  at  the  nodes  than  between 
them,  and  have  a  characteristically  ribbed  surface  upon  drying. 

The  leaves  of  the  Northern  Flints  are  long  and  slender  and  frequently  a 
light  green.  Those  of  the  Southern  Dents  are  proportionately  wider  and 
shorter  and  are  often  dark  green.  They  are  set  upon  culms  whose  internodes 
are  proportionately  longer  and  more  slender  in  the  Northern  Flints,  and  less 
prone  to  become  greatly  shortened  at  the  internodes  immediately  above  the 
ear. 

If  we  ignore  such  abnormalities  as  differences  in  carbohydrate  composition 
and  condensation,  these  two  races  of  Zea  mays  still  are  widely  different  from 
one  another — as  compared  to  differences  between  their  wild  relatives  in  the 
Andropogoneae  or  the  Maydeae.  The  differences  in  internode  pattern  and 
proportion  and  in  leaf  shape  are  similar  to  those  frequently  found  between 
species  of  the  same  genus.  The  differences  in  pedicellation  of  the  upper  spike- 
let  would  be  more  characteristic  of  genera  and  sub-genera.  On  the  other 
hand,  in  the  whorling  of  the  central  spike  (whorls  of  2  versus  whorls  of  3) 
is  the  kind  of  difference  which  would  ordinarily  separate  genera  or  even 
groups  of  genera.  On  a  par  with  this  difference  are  those  in  the  cupule  (the 
bony  cup  in  which  the  kernels  are  attached  in  pairs).  They  are  so  difficult  to 
observe  that  we  cannot  discuss  these  until  the  general  morphology  of  this 
organ  has  been  described.  If  we  sum  up  the  morphological  evidence,  it  is  clear 
that  the  fundamental  differences  between  the  Northern  Flints  and  the 
Southern  Dents  are  similar  to  those  which  differentiate  distantly  related 
species  (or  even  genera)  among  related  wild  grasses.  There  is  every  morpho- 
logical indication,  therefore,  that  we  are  dealing  with  two  fundamentally 
different  germ  plasms. 

The  cytological  facts  reported  above  lend  further  weight  to  the  conclusion 
that  the  Northern  Flints  are  basically  different  from  the  Southern  Dents. 
The  former  have  chromosomes  which  are  essentially  knobless  at  pachytene. 
The  latter  average  nearly  one  knob  per  chromosome  (Fig.  8.2).  Heterochro- 
matic  knobs  are  known  in  other  grasses  besides  Zea  mays.  In  these  other 
genera,  their  presence  or  absence,  from  such  evidence  as  is  available,  seems 
to  be  characteristic  of  whole  species  or  groups  of  species.  Such  a  difference 
between  the  Flints  and  Dents  indicates  that  we  are  dealing  with  two  funda- 
mentally different  germ  plasms.  It  has  been  shown  in  Guatemala  (Mangels- 
dorf  and  Cameron,  1942)  and  in  Mexico  (Anderson,  1946)  that  the  varieties 
with  many  knobs  are  morphologically  and  ecologically  different  from  those 
with  low  numbers  of  knobs. 

A  further  indication  that  these  two  germ  plasms  are  physiologically  dif- 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE  137 

ferent  is  given  by  their  pachytene  behavior.  The  pachytene  chromosomes  of 
the  Northern  Flints  are  easy  to  smear  and  give  sharp  fixation  images.  South- 
ern Dents  are  more  ditlficult  to  smear.  The  chromosomes  do  not  spread  out 
well  and  do  not  stain  sharply.  This  is  not  a  result  of  differences  in  knob  num- 
ber, since  some  of  the  Mexican  Dents  with  few  knobs  are  equally  difilicult  to 
smear.  Whatever  the  physiological  significance  of  this  reaction,  it  is  direct 
evidence  for  a  difference  in  the  chemistry  of  the  germ  cells.  Again  such  dif- 
ferences in  stainability  are  more  often  met  with,  between  genera,  than  they 
are  in  different  strains  of  the  same  species. 

There  is  genetic  evidence  for  the  difference  between  Southern  Dents  and 
Northern  Flints,  in  the  behavior  of  crosses  between  them.  The  F/s  are  fully 

TABLE  8.1 

PERCENTAGE  OF  STERILE  OR  BARREN  PLANTS  IN 

GOURDSEED,  LONGFELLOW,  AND  Fa  GENERATION 

OF  CROSS  GOURDSEED  X  LONGFELLOW 


Total 
Number 
of  Plants 


Gourdseed 

Longfellow 

Fo  Gourdseed X Longfellow 


Sterile 

Normal 

or 

Ear 

Barren 

37 

63 

2 

98 

52 

48 

fertile  and  exhibit  extreme  hybrid  vigor.  The  Fo's  show  a  high  percentage  of 
completely  barren  plants — plants  which  formed  ears  but  set  little  or  no  seeds, 
either  because  of  sterility  or  because  they  were  too  weak  to  mature  success- 
fully— and  plants  which  managed  to  set  seeds,  though  their  growth  habit 
indicates  fundamental  disharmonies  of  development. 

Table  8.1  shows  the  percentages  of  good  ears  and  plants  which  were  either 
without  ears  or  on  which  the  ears  had  failed  to  set  any  seed,  for  Gourdseed- 
Dent,  Longfellow  Flint,  and  their  F2,  when  grown  in  Iowa.  Like  Southern 
Dents  generally,  the  Gourdseed  is  less  adapted  to  central  Iowa  than  is  Long- 
fellow Flint.  An  F2  between  these  two  varieties,  however,  has  a  much  greater 
percentage  than  either  parent  of  plants  which  are  so  ill-adapted  that  they 
either  produce  no  visible  ear,  or  set  no  seed  if  an  ear  is  produced.  Similar 
results  were  obtained  in  other  crosses  between  Northern  Flints  and  Southern 
Dents,  both  in  Missouri  and  in  Iowa.  From  this  we  conclude  that  they  are 
so  genetically  different  from  one  another  that  a  high  percentage  of  their  F2 
recombinations  are  not  able  to  produce  seed,  even  when  the  plants  are  care- 
fully grown  and  given  individual  attention. 

SUMMARY 
The  common  dent  corns  of  the  United  States  Corn  Belt  were  created 
de  novo  by  American  farmers  and  plant  breeders  during  the  nineteenth  cen- 


138  EDGAR  ANDERSON  AND  WILLIAM  L  BROWN 

tury.  They  resulted  in  a  large  measure  from  deliberate  crossing  and  re- 
crossing  of  two  races  of  maize  (the  Northern  Flints  and  the  Southern  Dents) 
so  different  that,  were  they  wild  grasses,  they  would  be  considered  as  totally 
different  species  and  might  well  be  placed  in  different  genera.  The  origin  of 
two  so-different  races  within  cultivated  maize  is  an  even  larger  problem  and 
one  outside  the  scope  of  this  discussion.  It  may  be  pointed  out  parentheti- 
cally that  the  Tripsacum  hypothesis  (Mangelsdorf  and  Reeves,  1945)  would 
not  only  account  for  variation  of  this  magnitude,  it  would  even  explain  the 
actual  direction  of  the  difference  between  these  two  races  of  maize.  However, 
the  relation  between  maize  and  Tripsacum  on  any  hypothesis  is  certainly  a 
most  complicated  one  (Anderson,  1949).  It  would  be  more  effective  to  post- 
pone detailed  discussions  of  this  relationship  until  the  comparative  morphol- 
ogy of  the  inflorescences  of  maize  and  of  Tripsacum  is  far  better  understood 
than  it  is  at  present. 

SIGNIFICANCE  TO  MAIZE  BREEDING 

Derivation  of  the  commercial  field  corns  of  the  United  States  by  the  de- 
liberate mingling  of  Northern  Flints  and  Southern  Dents  is  a  fact.  Unfortu- 
nately, it  is  a  fact  which  had  passed  out  of  common  knowledge  before  the 
present  generation  of  maize  breeders  was  educated.  From  the  point  of  view 
of  practical  maize  breeding,  either  hybrid  or  open-pollinated,  it  is  of  central 
importance.  Briefly,  it  means  that  the  maize  germ  plasms  now  being  worked 
with  by  plant  breeders  are  not  varying  at  random.  They  are  strongly 
centered  about  two  main  centers  or  complexes.  Such  practical  problems  as 
the  development  and  maintenance  of  inbreds,  the  detection  of  combining  abil- 
ity, and  the  most  effective  utilization  of  hybrid  vigor  need  to  be  rethought 
from  this  point  of  view.  Detailed  experiments  to  provide  information  for  such 
practical  questions  already  are  well  under  way.  While  these  experiments  are 
not  yet  far  enough  along  to  give  definite  answers,  they  have  progressed  far 
enough  to  allow  us  to  speak  with  some  authority  on  these  matters. 

HETEROSIS 

The  heterosis  of  American  Corn  Belt  dents  acquires  a  new  significance  in 
the  light  of  these  results,  and  practical  suggestions  as  to  its  most  efficient 
utilization  take  on  a  new  direction.  We  are  immediately  led  to  the  hypothesis 
that  the  heterosis  we  are  working  with  is,  in  part  at  least,  the  heterosis  ac- 
quired by  mingling  the  germ  plasms  of  the  Northern  Flints  and  the  Southern 
Dents. 

Insofar  as  hybrid  vigor  is  concerned,  the  hybrid  corn  program  largely  has 
served  to  gather  some  of  the  dispersed  vigor  of  the  open-pollinated  dents. 
Preliminary  results  indicate  that  this  has  not  been  done  efliciently  in  terms 
of  what  might  be  accomplished  with  somewhat  more  orientation. 

The  early  days  of  the  hybrid  corn  program  were  dominated  by  the  hy- 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE  139 

pothesis  that  one  could  inbreed  this  vigorous  crop,  identify  the  inferior 
strains  in  it,  and  then  set  up  an  elite  cross-pollinated  germ  plasm.  This  hy- 
pothesis was  clearly  and  definitely  stated  by  East  and  Jones  {Inbreeding  and 
Outbreeding,  1919,  pp.  216-17). 

Experiments  with  maize  show  that  undesirable  qualities  are  brought  to  light  by  self- 
fertilization  which  either  eliminate  themselves  or  can  be  rejected  by  selection.  The  final  re- 
sult is  a  number  of  distinct  types  which  are  constant  and  uniform  and  able  to  persist  in- 
definitely. They  have  gone  through  a  process  of  purification  such  that  only  those  individu- 
als which  possess  much  of  the  best  that  was  in  the  variety  at  the  beginning  can  survive.  The 
characters  which  they  (pure  lines)  have,  can  now  be  estimated  more  nearly  at  their  true 
worth.  By  crossing,  the  best  qualities  which  have  been  distributed  to  the  several  inbred 
strains  can  be  gathered  together  again  and  a  new  variety  recreated.  After  the  most  desirable 
combinations  are  isolated,  their  recombination  into  a  new  and  better  variety,  which  could 
be  maintained  by  seed  propagation,  would  be  a  comparatively  easy  undertaking. 

Though  other  corn  breeders  and  corn  geneticists  may  not  have  committed 
themselves  so  definitely  in  print,  such  a  notion  was  once  almost  universal 
among  hybrid  corn  experts.  Modified  versions  of  it  still  influence  breeding 
programs  and  are  even  incorporated  in  elementary  courses  in  maize  breeding. 

The  facts  reported  above  would  lead  us  to  believe  that  heterosis,  having 
resulted  from  the  mingling  of  two  widely  different  germ  plasms,  will  probably 
have  many  genes  associated  with  characters  which  in  their  relatively  homo- 
zygous state  are  far  from  the  Corn  Belt  ideal  of  what  a  corn  plant  should  look 
like.  It  is  highly  probable  that  much  of  the  so-called  "junk"  revealed  by  in- 
breeding was  extreme  segregants  from  this  wide  cross,  and  that  it  was  closely 
associated  with  the  genes  which  gave  open-pollinated  dents  their  dispersed 
vigor.  It  is  significant  that  some  very  valuable  inbreds  (L317  is  a  typical  ex- 
ample) have  many  undesirable  features.  For  this  reason,  many  such  inbreds 
are  automatically  eliminated  even  before  reaching  the  testing  stage. 

If  one  accepts  the  fact  that  Corn  Belt  dents  resulted  from  the  compara- 
tively recent  mingling  of  two  extremely  different  races  of  maize,  then  on  the 
simplest  and  most  orthodox  genetic  hypotheses,  the  greatest  heterosis  could 
be  expected  to  result  from  crosses  between  inbreds  resembling  the  Southern 
Dents  and  inbreds  resembling  the  Northern  Flints.  If  heterosis  (as  its  name 
implies)  is  due  to  heterozygous  genes  or  segments,  then  with  Corn  Belt  corn 
on  the  whole  we  would  expect  to  find  the  greatest  number  of  differing  genes 
when  we  reassembled  two  inbreds — one  resembling  the  Northern  Flint,  the 
other  resembling  the  original  Southern  Dent. 

Theory  (Anderson,  1939a),  experiment  (Anderson,  1939b;  Brown,  1949), 
and  the  results  of  practical  breeding  show  that  linkage  systems  as  differenti- 
ated as  these  break  up  very  slowly.  On  the  whole,  the  genes  which  went  in 
together  with  the  Northern  Flints  still  tend  to  stay  together  as  we  have 
demonstrated  above.  This  would  suggest  that  in  selecting  inbreds,  far  from 
trying  to  eliminate  all  of  the  supposed  "junk,"  we  might  well  attempt  to 
breed  for  inbreds  which,  though  they  have  good  agronomic  characters  like 
stiffness  of  the  stalk,  nevertheless  resemble  Northern  Flints.  On  the  other 
hand,  we  should  breed  also  for  those  which  resemble  Southern  Dents  as  close- 


140  EDGAR  ANDERSON  AND  WILLIAM  L.  BROWN 

ly  as  they  can  and  still  be  relatively  easy  to  grow  and  to  harvest.  It  would 
seem  as  if  the  opposite  generally  has  been  done.  A  deliberate  attempt  has 
been  made  to  produce  inbreds  which  look  as  much  as  possible  like  good  Corn 
Belt  maize  in  spite  of  being  inbreds. 

There  are,  of  course,  practical  necessities  in  breeding.  In  this  direction  the 
work  of  corn  breeders  is  a  remarkable  achievement.  Strong  attention  to  lodg- 
ing resistance,  to  desirable  kernel  shapes  and  sizes,  and  to  resistance  to 
drought  and  disease  has  achieved  real  progress.  The  inbred-hybrid  method 
has  permitted  much  stronger  selection  for  these  necessary  characters  than 
was  possible  with  open-pollinated  maize.  Most  Corn  Belt  dents  now  plant 
well,  stand  well,  and  harvest  well. 

Perhaps  partly  because  of  these  practical  points  there  has  been  a  conscious 
and  unconscious  attempt  on  the  part  of  many  breeders  to  select  for  inbreds 
which  are  like  the  Corn  Belt  ideal  in  all  characters,  trivial  and  practical 
alike.  The  corn  shows  are  now  out-moded,  but  corn  show  ideals  still  influence 
corn  breeding.  For  instance,  there  has  been  an  effort  to  produce  plants  with 
greatly  arching  leaves,  whose  margins  are  uniformly  ruffled.  Such  characters 
are  certainly  of  a  trivial  nature  and  of  secondary  importance  in  practical  pro- 
grams. Any  potential  heterosis  closely  associated  with  upright  leaves,  yellow 
green  leaves,  tillering,  or  blades  on  the  husk  leaves  has  seldom  had  a  chance 
to  get  into  inbreds  where  it  could  be  tested  on  a  basis  of  achievement.  It 
would  seem  highly  probable  that,  in  not  basing  the  selection  of  inbreds  more 
soundly  on  performance,  we  have  let  much  potential  heterosis  slip  through 
our  sieve  of  selection. 

Heterosis  Reserves 

These  considerations  lead  us  to  believe  that  there  is  probably  a  good  deal 
of  useful  heterozygosis  still  ungathered  in  high  yielding  open-pollinated 
varieties.  There  is  also  a  distinct  possibility  that  still  more  could  be  added 
by  going  back  to  the  Northern  Flints  and  Southern  Dents  with  the  specific 
object  of  bringing  in  maximum  heterozygosity.  From  our  experience  it  is 
more  likely  that  superior  heterosis  is  to  be  found  among  the  best  flints  than 
among  the  best  dents.  On  the  whole,  the  Northern  Flints  have  been  farthest 
from  the  corn  breeders'  notion  of  what  a  good  corn  plant  should  look  like. 
Flint-like  characteristics  (tillering,  for  example)  have  been  most  strongly 
selected  against,  both  in  the  open-pollinated  varieties  and  the  inbreds  derived 
from  them. 

Several  of  the  widely  recognized  sources  of  good  combining  inbreds  are 
open-pollinated  varieties  with  a  stronger  infusion  of  Northern  Flints  than 
was  general  in  the  Corn  Belt.  This  is  particularly  true  of  Lancaster  Surecrop, 
the  excellence  of  whose  inbreds  was  early  recognized  by  several  breeders  in 
the  United  States  Department  of  Agriculture.  In  our  opinion,  it  is  probable 
that  the  greater  proportion  of  flint  germ  plasm  in  Lancaster  Surecrop  has 


ORIGIN  AND  SIGNIFICANCE  OF  CORN   BELT  MAIZE  141 

made  it  an  outstandin<f  source  of  inbreds  of  proven  highly  sj)ecific  combining 
ability  when  used  with  other  Corn  Belt  inbreds.  This  is  not  an  isolated  ex- 
ample, and  even  more  extreme  cases  could  be  cited.  We  think  it  is  a  reason- 
able working  hypothesis  that  Northern  Flint  varieties  of  superior  productiv- 
ity might  be  efficient  sources  of  improved  heterozygosity  for  the  United 
States  Corn  Belt. 

Morphological  Characters  as  Related  to  Heterosis 

To  put  this  hypothesis  in  different  language,  morphological  characters,  if 
carefully  chosen,  may  be  used  as  criteria  of  specific  combining  ability  in  Corn 
Belt  inbreds.  Before  presenting  data  bearing  directly  on  this  hypothesis,  two 
points  need  to  be  emphasized  and  discussed:  (1)  the  effective  selection  of 
morphological  criteria,  and  (2)  the  relativity  of  all  measures  of  effective 
combining  ability. 

Previous  studies  (Kiesselbach,  1922;  Jenkins,  1929;  and  others)  have  indi- 
cated that  the  only  positive  correlations  between  the  morphology  of  inbreds 
and  their  combining  ability  are  those  involving  characters  of  the  inbreds 
which  are  indicative  of  plant  vigor.  Reference  to  these  investigations  shows 
that  the  characters  chosen  were  such  superficial  measurements  as  date  of 
silking  and  tasseling,  plant  height,  number  of  nodes,  number  of  ears,  ear 
diameter,  etc.  Unfortunately,  the  morphology  of  the  maize  plant  is  not  a 
simple  matter.  It  is  so  complex  that  one  needs  technical  help  on  morphology 
quite  as  much  as  he  would  in  biochemistry  were  he  studying  the  concentra- 
tions of  amino  acids  in  the  developing  kernel. 

Accordingly,  we  first  familiarized  ourselves  thoroughly  with  the  technical 
agrostological  facts  concerning  the  detailed  gross  morphology  of  grasses  in 
general  and  Zea  in  particular.  Just  as  in  the  case  of  a  biochemical  study  of  the 
kernel,  we  found  that  further  original  research  was  necessary  if  the  investiga- 
tion was  to  be  carried  on  effectively.  We  have  accordingly  undertaken  de- 
tailed studies  of  internode  patterns  and  branching  of  the  inflorescence;  the 
venation,  size,  and  shape  of  the  male  spikelet,  the  development  of  the  husk 
leaf  blades,  the  external  anatomy  of  the  cob,  and  the  morphology  of  the 
shank.  Some  of  these  investigations  are  still  continuing,  and  must  continue 
if  inbred  morphology  and  combining  ability  are  to  be  effectively  correlated. 

It  is  impossible  to  produce  an  absolute  measure  of  combining  ability. 
When  one  speaks  of  combining  ability  of  two  inbreds,  he  always  refers  to 
their  behavior  with  each  other  compared  to  their  behavior  with  certain  other 
inbreds  or  open-pollinated  varieties.  This  is  such  a  relative  measure  that  the 
scoring  of  a  particular  Fi  cross  as  very  low  or  very  high  in  combining  ability 
might  depend  solely  upon  our  previous  experience  with  the  two  inbreds.  We 
may  illustrate  this  point  with  an  extreme  example.  Let  us  suppose  that  we 
have  inbreds  IF  and  2F  derived  directly  from  Northern  Flints,  and  inbreds 
lOD  and  IID  derived  from  Southern  Dents.  Were  we  to  cross  IF  X  2F  and 


142  EDGAR  ANDERSON  AND  WILLIAM  L.  BROWN 

lOD  X  IID  we  would  expect  relatively  little  heterosis  within  either  of  the 
crosses.  Accordingly,  when  we  crossed  2F  X  IID  we  would  rate  this  cross 
as  having  high  specific  combining  ability.  On  the  other  hand,  had  we  origi- 
nally crossed  2F  X  lOD  and  IID  X  IF,  then  there  would  probably  have 
been  almost  equally  great  heterosis  in  each  of  the  crosses.  Had  these  been 
used  as  a  basis  for  comparing  the  heterosis  of  2F  X  IID,  then  our  notion  as 
to  the  amount  of  heterosis  in  these  crosses  would  have  been  very  different 
than  it  would  have  been  had  comparisons  been  made  with  IF  X  2F  or 
lOD  X  IID. 

If  the  germ  plasms  of  the  two  main  races  of  maize  involved  in  Corn  Belt 
dents  are  still  partially  intact  as  a  result  of  linkages,  it  should  be  possible  to 
classify  inbreds  on  the  basis  of  morphological  differences  according  to  their 
flint  and  dent  tendencies.  If  this  can  be  done,  and  if  genetic  diversity  is  im- 
portant in  bringing  about  a  heterotic  effect  in  hybrids,  one  should  be  able  to 
predict  with  some  accuracy  the  relative  degree  of  heterosis  to  be  expected 
from  crossing  any  two  inbred  lines.  With  this  hypothesis  as  a  background,  a 
series  of  experiments  was  started  three  years  ago  to  determine  whether  or  not 
hybrid  vigor  in  maize,  as  expressed  in  terms  of  grain  yield,  could  be  predicted 
on  the  basis  of  morphological  differences  of  inbreds  making  up  the  Fi  hybrids. 

Fifty-six  relatively  homozygous  inbred  lines  consisting  of  eighteen 
U.S.D.A.  or  experiment  station  lines,  and  thirty-eight  strains  developed  by 
the  Pioneer  Hi-Bred  Corn  Company  were  scored  for  the  following  character- 
istics: row  number,  kernel  length,  denting,  development  of  husk  leaf  blades, 
number  of  secondary  tassel  branches,  glume  length,  and  chromosome  knob 
number.  For  each  of  these  characteristics  the  two  extremes  in  the  eastern 
United  States  are  to  be  found  in  Southern  Dents  and  Northeastern  Flints. 
At  least  twelve  plants  of  each  of  the  fifty-six  inbreds  were  scored,  and  these 
scores  were  then  averaged  to  give  a  mean  value  for  the  line.  The  resulting 
means  were  translated  into  numerical  index  values,  in  which  a  low  value 
represents  Northern  Flint-like  tendencies,  and  a  high  value  Southern  Dent- 
like tendencies.  For  example,  the  mean  row  number  values  for  the  inbreds 
studied  ranged  from  11.2  to  19.5.  These  were  arranged  in  the  following  index 
classes. 

1  2  3  4  S  6  7 

11.2-11.7     11.8^12.3     12.4-12.9     13.0-13.5     13.6-14.1     14.2-14.7     14.8-15.3 

8  9  10  11  12  13  14 

15.4-15.9     16.0-16.5     16.6-17.1     17.2-17.7     17.8-18.3     18.4-18.9     19.0-19.5 

Index  values  for  the  other  characteristics  were  arranged  similarly,  and 
from  the  individual  characteristic  inbred  indices  (each  being  given  equal 
weight)  a  total  "Inbred  Index"  was  determined  as  is  shown  by  example  in 
Table  8.2. 

After  index  values  had  been  determined  for  the  inbreds,  single  cross  combi- 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE 


143 


nations  were  made  and  these  tested  for  yield.  In  1948,  sixty-six  single  crosses 
were  grown  in  yield  tests  in  Iowa  and  in  Illinois.  Each  Fi  hybrid  was  repli- 
cated six  times  in  each  test.  At  the  end  of  the  season,  yield  of  grain  was  de- 
termined on  the  basis  of  15  per  cent  moisture  corn.  Actual  yields  in  bushels 
per  acre  and  morphological  differences  of  the  inbreds  involved  in  each  of  the 
crosses  were  then  plotted  on  a  scatter  diagram  as  shown  in  Figure  8.6.  It  will 
be  noted  that  although  the  observations  exhibit  considerable  scatter,  there  is 
a  tendency  for  grain  yields  in  single  crosses  to  increase  as  the  morphological 
differences  between  the  inbreds  making  up  the  crosses  become  greater. 
Actually  the  correlation  coefficient  between  yield  and  index  differences  in 
this  case  was  r  =  +.39. 

The  experiment  was  continued  in  1949,  in  which  100  Fi  hybrids  were 
tested  for  yield.  In  this  experiment  three  characters  only  were  used  to  deter- 

TABLE  8.2 
INBRED  INDICES  BASED  ON  SEVEN  CHARACTERS 


Inbreds 

Row 

No. 

Kernel 
Length 

Dent- 
ing 

Husk 
Leaves 

Tassel 
Branches 

Spikelet 
Length 

Chromo- 
some 
Knobs 

^  ,      J      Sums  of  7 
Inbred     ^.„ 
^    ,          Differences 
Index       .,,      ^  „. 

without  Signs 

Hy 

Oh40b.. 
MYl... 

9 

2 

14 

14 

8 

11 

4 

4 
14 

14 

1 
14 

5 

4 

14 

6 
1 
6 

12 

3 
9 

f± 

IV/           30 

82          / 

mine  the  index  of  relationship  between  the  inbreds  used.  These  were  row 
number,  kernel  length,  and  degree  of  development  of  husk  leaf  blades. 
Elimination  in  this  experiment  of  certain  morphological  characteristics  used 
previously  was  done  largely  to  facilitate  ease  and  speed  of  scoring.  It  had 
been  determined  previously  that,  of  the  several  characteristics  used,  those 
having  the  highest  correlation  with  yield  were  differences  in  row  number, 
kernel  length,  and  husk  leaf  blades.  There  was  likewise  known  to  be  a  rather 
strong  association  between  each  of  these  characteristics  and  tassel  branch 
number,  denting,  glume  length,  internode  pattern,  and  chromosome  knob 
number.  Therefore  the  scoring  of  these  three  characteristics  probably  covers 
indirectly  nearly  as  large  a  segment  of  the  germ  plasm  as  would  scores  based 
on  all  seven  characteristics. 

The  1949  tests  in  which  each  entry  was  replicated  six  times  in  each  loca- 
tion were  again  grown  both  in  Iowa  and  Illinois.  Yields  from  these  tests, 
plotted  against  index  differences  of  the  inbreds,  are  shown  in  Figure  8.7. 
As  in  the  previous  year's  data,  a  pronounced  tendency  was  shown  for  hybrids 
made  up  of  inbreds  of  diverse  morphology  to  produce  higher  grain  yields  than 
hybrids  consisting  of  morphologically  similar  inbreds.  The  correlation  co- 


1948 


. 

; 

# 

• 

115 

" 

• 

• 

• 

u 

\ 

Mk 

• 

W 

• 

V 

DC 

no 

. 

• 

• 

# 

(J 

• 

• 

• 

• 

• 

• 

• 
• 

• 

# 

^ 

• 

• 

• 

• 

• 

D 

105 

• 

• 

• 

• 
• 

• 

• 

• 
• 

• 

• 

• 

OQ 

•  • 

• 

• 

• 

• 

• 

1 

• 
• 

• 
• 

• 

# 

• 

Q 

100 

- 

• 

• 

• 

• 

J 

•  • 

• 

• 

• 

LiJ 

•  # 

• 

> 

95 
90 

• 

• 

85 

• 

0  5  10 

NDEX  DIFFERENCE 


r=+.38 


Fig.  8.6 — Scatter  diagram  depicting  relationship  between  grain  yields  of  66  single  cross 
hybrids  and  morphological  differences  of  inbred  parents  of  the  hybrids.  Explanation  in  text. 


ORIGIN   AND  SIGNIFICANCE  OF  CORN   BELT  MAIZE 


145 


efficient  between  yield  and  index  differences  is  r  =  +.40,  a  significant  value 
statistically. 

In  terms  of  practical  corn  breeding,  the  distribution  of  single  crosses  in 
Figures  8.6  and  cS.7  is  of  particular  significance.  If  these  observations  are 
critical  (we  have  produced  a  repea table  result)  it  means  that  one  could  have 
eliminated  from  the  testing  program  the  lower  one-third  of  the  crosses  on  the 
basis  of  index  differences,  without  losing  any  of  the  top  10  per  cent  of  the 
highest  yielding  hybrids.  In  the  case  of  the  100  hybrids  in  Figure  8.7,  one  could 
have  eliminated  from  testing  35  per  cent  of  the  crosses,  thereby  permitting  the 
inclusion  of  35  additional  hybrids  in  this  particular  testing  area.  If  further 

1949 


130 


125  • 


120  ■ 


3  lis 
CD 

I     I  10 

a 

LJ  10^ 


LJ 

q: 
u 


> 


00 


95 


•     •  •  • 


0 


•     •  •  •  • 


:.•  • 


•  • 


•    • 


5  10  15  20 

INDEX  DIFFERENCE 


25 


r=  +.40 


^x 


Fig.  8.7 — Scatter  diagram  depicting  relationship  between  grain  yields  of  100  single  cross 
hybrids  and  morphological  differences  of  inbred  paren  ts  of  the  hybrids.  Explanation  in  text . 


146  EDGAR  ANDERSON  AND  WILLIAM  L  BROWN 

experiments  show  that  the  method  is  reliable,  such  a  procedure  should 
expedite  most  corn  breeding  programs. 

Our  method  of  scoring  does  not  take  into  account  the  variation  brought 
about  by  the  infusion  of  germ  plasm  other  than  that  from  Northern  Flints 
and  Southern  Dents.  Perhaps  this  is  one  reason  why  we  have  not  ob- 
tained higher  correlations  between  differences  in  inbred  morphology  and 
yield.  There  are  a  few  inbreds  in  the  Corn  Belt  which  appear  to  be  affiliated 
either  with  Caribbean  flints  or  the  Basketmaker  complex.  Scoring  of  such 
inbreds  on  a  scale  designed  for  Northern  Flints  and  Southern  Dents  un- 
doubtedly leads  to  conflicting  results.  It  is  hoped  that  experiments  now  in 
progress  will  aid  in  clarifying  this  situation. 

SIGNIFICANCE  OF  FLINT-DENT  ANCESTRY 
IN  CORN   BREEDING 

The  Flint-Dent  ancestry  of  Corn  Belt  maize  bears  upon  many  other  breed- 
ing problems  besides  those  concerned  with  heterosis.  Its  widest  usefulness  is 
in  giving  a  frame  of  reference  for  observing  and  thinking  about  the  manifold 
and  confusing  variation  of  Corn  Belt  maize.  When  one  becomes  interested 
in  any  particular  character  of  the  corn  plant,  he  no  longer  needs  to  examine 
large  numbers  of  inbreds  to  understand  its  range  of  variation  and  its  general 
over-all  direction.  He  merely  needs  to  examine  a  few  inbreds,  and  a  Northern 
Flint  and  a  Southern  Dent.  A  good  part  of  the  variation  will  then  be  seen  to 
fall  into  a  relatively  simple  series  from  an  extreme  Northern  Flint  type  to  the 
opposite  Southern  Dent  extreme,  with  various  intermediates  and  recombina- 
tions in  between.  This  is  quite  as  true  for  physiological  or  biochemical  char- 
acters as  for  glumes,  lemmas,  or  other  morphological  characters.  One  is  then 
ready  to  study  further  inbreds  with  a  framework  in  his  mind  for  sorting  out 
and  remembering  the  variation  which  he  finds. 

The  actual  breeding  plot  efficiency  of  this  understanding  will  be  clearer  if 
we  cite  a  practical  example.  Now  that  corn  is  picked  mechanically,  the  size, 
shape,  texture,  and  strength  of  shank  are  important.  When  maize  was  picked 
by  hand,  the  hand  had  a  brain  behind  it.  Variations  in  ear  height,  in  the 
stance  of  the  ear,  and  in  the  strength  and  shape  of  the  shank  were  of  minor 
significance.  Now  that  machines  do  the  work,  it  is  of  the  utmost  practical  im- 
portance to  have  the  shank  standardized  to  a  type  adapted  to  machine 
harvesting.  When  this  necessity  was  brought  to  our  attention  a  few  years 
ago,  there  were  few  published  facts  relating  to  variation  in  the  shank.  Exam- 
ination of  a  few  inbreds  showed  that  though  this  organ  varied  somewhat 
within  inbreds,  it  varied  more  from  one  line  to  another  than  almost  any 
other  simple  feature  of  the  plant.  We  accordingly  harvested  typical  shanks 
from  each  of  164  inbreds  being  grown  for  observation  in  the  breeding  plots 
of  the  Pioneer  Hi-Bred  Corn  Company.  We  also  examined  a  number  of 
Northern  Flints,  and  had  they  been  available,  we  would  have  studied  the 


ORIGIN  AND  SIGNIFICANCE  OF  CORN  BELT  MAIZE 


147 


shanks  on  typical  Southern  Dents.  However,  simply  b}-  using  the  hyj)othesis 
that  one  extreme  would  have  to  come  in  from  the  Northern  Flints,  the  other 
from  the  Southern  Dents,  we  were  able  within  one  working  day  to  tabulate 
measurable  features  of  these  shanks  and  to  incorporate  all  the  facts  in  a 


< 


X 
h- 

o 

z 

LJ 


< 

I- 

o 

I- 


'4 

4 

V 

)l 

i 

V 

4 

265 

i               V 
4          4 

4     4 

U           4 

4 

'^4    4 

V 

u 

'i 

4 

V 

t 

210 

4     '*    ^      4 

V     / 
^4 

4 

\ 

4 

/ 

U     4 

'i 

u 

w 

4 

)r 

W' 

4^4  ^i  4  ^i 

4 

u 

% 

^w 

135 

i,  ^ 

V 

V 

V 

*- 

V 

^ 

¥ 

t         )i 

^        4 

i 

a 
^ 

4 

• 

60 

10 


15 


20 


WIDTH  OF  MID-INTERNODE  OF  SHANK 


LENGTH   OF 

MAXIMUM   WIDTH - 

CONDENSED 

LONGEST  INTERNODE 

MINIMUM    WIDTH 

INTERNODES 

•   0-15 

•  0 

\o 

¥  16-35 

i  1-3 

■*  1-2 

436-  + 

i4-  + 

•  3■-^ 

Fig.  8.8 — Pictorialized  diagram  showing  relationship  in  164  Corn  Belt  inbreds  of  the  fol- 
lowing shank  characters:  total  length,  width  of  mid  internode,  length  of  longest  internode, 
maximum  width  minus  minimum  width,  and  number  of  condensed  internodes. 


148  EDGAR  ANDERSON  AND  WILLIAM  L.  BROWN 

pictorialized  scatter  diagram  (Fig.  8.8).  Using  the  method  of  Extrapolated 
Correlates  we  were  able  to  reconstruct  the  probable  shank  type  of  the 
Southern  Dents.  (We  later  grew  and  examined  them  and  verified  our  predic- 
tions.) We  arranged  most  of  the  facts  concerning  variation  in  shank  type  in 
United  States  inbreds  in  a  single,  easily  grasped  diagram.  All  the  technical  in- 
formation needed  as  a  background  for  breeding  was  available  after  two  days' 
work  by  two  people.  Without  the  Northern  Flint-Southern  Dent  frame  of 
reference  for  these  miscellaneous  facts,  we  might  have  worked  around  the 
problem  for  several  breeding  seasons  before  comprehending  this  general,  over- 
all picture. 

SUMMARY 

1.  Archaeological  and  historical  evidence  shows  that  the  common  dent 
corns  of  the  United  States  Corn  Belt  originated  mainly  from  the  purposeful 
mixing  of  the  Northern  Flints  and  the  Southern  Dents. 

2.  Cytological  and  genetic  evidence  point  in  the  same  direction  and  were 
used  in  the  earlier  stages  of  our  investigations  before  the  complete  historical 
evidence  had  been  located. 

3.  The  Northern  Flints  and  Southern  Dents  belong  to  races  of  maize  so 
dififerent  that,  were  they  wild  grasses,  they  would  certainly  be  assigned  to 
different  species  and  perhaps  to  different  genera.  Such  cytological  and 
genetical  evidence  as  is  available  is  in  accord  with  this  conclusion. 

4.  The  significance  of  these  facts  to  maize  breeding  problems  is  outlined. 
In  the  light  of  this  information,  the  heterosis  of  Corn  Belt  maize  would  seem 
to  be  largely  the  heterosis  acquired  by  mingling  the  germ  plasms  of  the 
Northern  Flints  and  Southern  Dents.  It  is  pointed  out  that  most  breeding 
programs  have  been  so  oriented  as  to  be  inefficient  in  assembling  the  dis- 
persed heterosis  of  the  open-pollinated  varieties  of  the  Corn  Belt.  The  possi- 
bility of  gathering  more  heterosis  from  the  same  sources  is  discussed  and  it  is 
suggested  that  more  might  be  obtained,  particularly  among  the  Northern 
Flints. 

5.  Morphological  characters  of  dents  and  flints,  if  carefully  chosen,  should 
be  useful  criteria  for  specific  combining  ability.  The  problem  of  selecting 
such  characters  is  described.  Two  seasons'  results  in  correlating  combining 
ability  and  flint-dent  differences  are  reported.  They  are  shown  to  be  statisti- 
cally significant  and  of  probable  practical  importance. 

6.  The  practical  advantages  of  understanding  the  flint-dent  ancestry  of 
Corn  Belt  maize  are  discussed  and  illustrated  by  example.  In  brief  these  facts 
provide  a  "frame  of  reference"  for  detecting,  organizing,  and  understanding 
much  of  the  manifold  variability  in  Corn  Belt  maize. 


ADRIANO  A.  BUZZATI-TRAVERSO 

Universiia,  Isfituto  di  Gene/ico,  Pavia,  Ifaly 


Chapter  9 

Heterosis  in 
Population  Genetics 


Population  genetics  is  the  study  of  the  genetic  structure  of  populations. 
Such  a  statement  may  look  at  first  to  be  a  truism,  a  tautology.  The  subject 
matter  of  our  research  becomes  very  intricate,  however,  as  soon  as  we  try  to 
specify  what  we  mean  by  the  above  definition.  The  terms  "genetic  structure" 
and  "population"  may  have  different  meanings  according  to  what  we  are 
willing  to  indicate  by  such  words.  It  therefore  seems  convenient  to  start 
with  an  analysis  of  the  terms  we  are  using.  Such  discussion  will  give  us  a 
chance  to  see  how  the  problem  of  heterosis  is  intimately  connected  with  the 
general  theme  of  population-genetical  studies.  A  few  experimental  data  will 
be  used  to  illustrate  such  points. 

Let  us  consider  first  what  we  mean  by  population.  If  we  take  a  dictionary 
definition,  we  find  in  Webster's  that  population  is  "all  the  people  or  in- 
habitants in  a  country  or  section."  It  means,  in  this  sense,  the  sum  of  indi- 
viduals present  at  a  certain  moment  over  a  more  or  less  arbitrarily  limited 
territory.  But  this  definition  does  not  correspond  to  the  requirements  of  our 
studies,  as  I  have  tried  to  show  elsewhere  (Buzzati-Traverso,  1950).  Such  a 
definition  is  a  static  one,  while  the  population,  as  considered  in  the  field  of 
population  genetics,  is  a  dynamic  concept.  We  are  interested  not  in  the 
number  of  individuals  present  at  a  certain  time  in  a  certain  place  and  their 
morphological  and  physiological  characteristics.  Instead,  we  are  concerned 
with  the  underlying  mechanisms  which  bring  about  such  characteristics,  and 
the  particular  size  the  population  reaches  at  any  particular  moment.  Since 
such  mechanisms  depend  upon  the  numerical  dynamics  of  the  population 
and  upon  heredity,  it  follows  that  our  concept  of  population  is  typically 
dynamic.  On  this  view,  then,  a  population  is  an  array  of  interbreeding  indi- 
viduals, continuous  along  the  time  coordinate. 

149 


150  ADRIANO  A.  BUZZATI-TRAVERSO 

Consideration  of  a  population  as  a  phenomenon  continuously  occurring  in 
time  makes  it  impossible  for  the  experimental  student  of  population  genetics 
to  get  a  direct  and  complete  picture  of  what  is  occurring  within  a  population 
at  any  particular  moment.  We  can  attempt  to  collect  data  on  the  population 
under  study  only  by  freezing  such  flowing  processes  at  particular  time  in- 
tervals. Collecting  observations  on  a  population  at  different  times  gives  us 
a  chance  to  extrapolate  the  direction  and  rate  of  the  processes  that  have 
occurred  within  the  population  during  the  time  elapsed  between  two  succes- 
sive sets  of  observations.  If  the  samples  studied  are  large  enough  and  give  an 
unbiased  picture  of  the  total  population  at  the  time  when  the  sample  is  being 
drawn,  this  experimental  procedure  ma}^  give  us  a  fairly  adequate  idea  of 
what  is  going  on  within  the  array  of  interbreeding  individuals  continuous 
along  the  time  coordinate.  That  sum  of  individuals  at  a  definite  time,  which 
one  usually  means  by  population,  is  of  interest  to  the  population  geneticist 
only  as  an  index  of  the  particular  evolutionary  stage  reached  by  the  array  of 
interbreeding  individuals.  Since  there  are  actual  breeding  and  genetic  rela- 
tionships between  the  individuals  of  any  such  array,  of  any  such  population, 
the  population  can  be  considered  as  the  natural  unit  of  our  studies. 

If  we  consider  now  what  we  mean  by  "genetic  structure,"  our  task  be- 
comes much  more  complex.  At  first  we  could  assume  that  the  genetic  struc- 
ture of  a  population  could  be  properly  described  in  terms  of  the  gene  frequen- 
cies present  at  a  certain  time  within  a  population.  But  this  is  only  part  of  the 
picture. 

For  the  total  description  of  the  genetic  structure  of  a  population  we  have 
to  consider  not  only  the  frequencies  of  existing  genes,  but  how  these  are 
fitted  within  the  chromosomes,  how  these  allow  the  release  of  variability  by 
means  of  recombinations,  how  large  is  the  amount  of  new  variability  pro- 
duced by  mutations,  and  several  other  factors  which  we  cannot  analyze  now. 
In  a  few  words,  the  study  of  population  genetics  aims  at  the  knowledge  of  the 
breeding  system  of  populations.  This,  as  we  shall  see,  is  a  rather  difficult  task 
because  of  the  complexity  of  factors  responsible  for  the  origin  and  evolution 
of  such  systems. 

EVOLUTIONARY  FACTORS  INVOLVED 

When  we  take  into  consideration  a  species  or  a  natural  population  at  a 
certain  stage,  we  have  to  assume  that  such  a  natural  entity  is  the  product  of 
a  series  of  evolutionary  factors  that  have  been  at  work  in  previous  times  and 
that  some,  or  all  of  them,  are  still  operating  on  the  population  while  we  are 
studying  it.  This  means  that  we  should  try  to  explain  the  genetic  structure  of 
the  population  in  terms  of  such  evolutionary  factors. 

Now,  if  we  are  willing  to  examine  the  nature  of  the  known  evolutionary 
agencies,  we  conclude  that  these  can  be  classified  into  two  types.  On  one  side 
we  find,  in  sexually  reproducing  organisms,  a  limited  number  of  chromo- 


HETEROSIS  IN   POPULATION   GENETICS  151 

somes,  linkage  between  genes,  sterility  mechanisms,  mating  discriminations, 
devices  favoring  inbreeding,  and  other  conservative  forces  that  aim  at  the 
preservation  of  certain  constellations  of  genes  over  a  large  number  of  genera- 
tions. On  the  other  side  we  find  mutation  pressure,  recombination  between 
chromosomes,  recombination  among  genes  due  to  crossing  over,  outbreeding 
devices,  migration  pressure,  and  other  revolutionary  forces  that  aim  at  the 
production  of  genetic  novelt3^ 

It  seems  reasonable  to  maintain  that,  at  any  particular  time,  a  species  or 
a  natural  population  can  be  considered  as  a  sort  of  compromise  between  the 
two  conflicting  forces — a  compromise  that  is  brought  about  through  the 
action  of  natural  selection.  In  other  words,  the  fine  adjustment  or  adaptation 
of  a  population  to  its  environment  is  the  expression  of  such  compromise.  At 
any  particular  time  the  terms  of  the  compromise  between  the  conflicting 
forces  are  always  different  as  compared  to  other  moments,  as  the  compro- 
mise itself  is  a  dynamic  process. 

In  order  to  reach  the  highest  possible  level  of  adaptation  with  respect  to  a 
certain  set  of  environmental  conditions,  natural  selection  is  discriminating  not 
only  for  or  against  a  certain  individual  genetic  constitution,  but  for  or  against 
a  group  of  individuals,  as  w-ell.  Sometimes  selection  acts  at  the  level  of  the  in- 
dividual, sometimes  it  operates  at  some  higher  level.  If  we  consider  a  genotype 
that  insures  resistance  against  an  infectious  disease,  present  in  a  certain  area  of 
distribution  of  a  species,  it  wall  be  obvious  that  an  individual  carrying  it  shall 
directly  benefit  by  it.  But  if  we  consider  a  genotype  producing  fecundity 
higher  than  the  average  of  the  population,  this  will  be  selected  by  the  mere 
fact  that  a  larger  number  of  individuals  having  such  genetic  constitution  will 
be  present  in  the  next  generation.  These,  in  their  turn,  shall  have  a  chance  of 
being  represented  in  the  next  generation  greater  than  that  of  individuals 
having  a  less  fertile  genotype.  The  individual  itself,  though,  obtains  no  direct 
advantage  from  such  selection. 

The  next  extreme  condition  we  can  consider  is  the  one  occurring  when  the 
advantage  of  the  individual  is  in  conflict  with  the  advantage  of  the  group. 
This  is  the  case,  for  instance,  of  a  genotype  that  would  extend  the  span  of 
life  far  beyond  the  period  of  sexual  activity — or  higher  fertility  linked  with 
antisocial  attitudes  in  the  case  of  man.  In  both  cases,  natural  selection  favor- 
ing the  preservation  of  the  group  will  discriminate  against  the  individual.  A 
similar  mechanism  must  have  played  a  great  role  in  various  critical  periods 
of  organic  evolution.  When  intergroup  selective  pressure  is  in  the  opposite 
direction  from  intragroup  selection,  a  sort  of  compromise  has  to  be  reached 
between  the  two  conflicting  tendencies.  This  can  be  reached  in  many  differ- 
ent ways  that  are  best  illustrated  by  the  great  variety  of  life  histories  and 
mating  systems  to  be  found  in  the  living  world. 

Those  factors  which  we  have  classified  as  conservative  tend  to  j)roduce 
genetic  homogeneity,  or  what  is  technically  known  as  homozygosis.  Factors 


152  ADRIANO  A.  BUZZATI-TRAVERSO 

that  we  have  named  revolutionary  tend  to  produce  genetic  heterogeneity  or 
heterozygosis.  Thus  we  come  to  the  conclusion  that  the  mentioned  compro- 
mise brought  about  by  selection  consists  of  the  pursuit  of  an  optimum  level 
of  hybridity  with  respect  to  the  conditions  under  which  the  organism  lives. 
Such  a  hybridity  optimum  is  the  product,  not  only  of  the  mutation  rate  and 
selective  value  of  single  genes,  but  also  depends  largely  upon  the  genetic  sys- 
tem and  the  mating  system — the  breeding  system — of  the  considered  species 
or  population. 

The  genetic  structure  of  natural  populations  cannot  be  solved  only  in 
terms  of  individual  variations  observable  in  the  group.  Instead,  it  must  be 
integrated  into  a  unitary  research  on  changes  in  gene  frequencies  as  related 
to  the  underlying  breeding  systems.  This  is  why  we  are  justified  in  consider- 
ing the  natural  population  as  a  unit,  since  individual  variations  must  be 
referred  to  the  genetic  balance  of  the  whole  aggregate  of  individuals. 

What  is  that  hybridity  optimum  I  was  speaking  about  but  heterosis?  How 
else  could  heterosis  be  defined  in  population  problems  other  than  that  type 
and  amount  of  heterozygosity  that  gives  the  population  or  the  individual  the 
best  adaptive  value  with  respect  to  the  conditions  in  which  the  organism 
lives?  With  this  view,  then,  it  becomes  feasible  to  analyze  experimentally 
what  morphological  and  physiological  characteristics  of  the  hybrids  produce 
the  better  adaptation. 

MECHANISMS  WHICH  PROMOTE  HYBRIDITY 

In  studying  how  heterosis  mechanisms  are  brought  about  in  living  crea- 
tures, we  may  attempt  a  sort  of  classification  of  the  devices  present  in  plants 
and  animals  insuring  hybridity.  Starting  from  the  most  complex  and  proceed- 
ing to  the  less  complex  cases,  we  can  distinguish  three  types  of  mechanisms: 
(1)  mating  systems,  (2)  chromosome  mechanisms,  and  (3)  gene  effects. 

We  will  not  discuss  in  detail  all  the  devices  insuring  hybridity  found  in 
plants  and  animals.  We  will  mention  a  few,  in  order  to  show  how  many  differ- 
ent paths  have  been  followed  in  evolution  to  reach  the  same  sort  of  results. 

Under  the  heading  "mating  systems"  we  may  mention  homo-  and  hetero- 
thally  among  fungi;  monoecism  and  dioecism,  incompatibility  mechanisms, 
and  heterostyly  among  flowering  plants.  Here,  in  some  cases  such  as  Primula 
scofica,  there  is  close  relation  between  the  variability  of  ecological  conditions, 
and,  therefore,  of  selection  pressure  and  the  efficiency  of  the  incompatibility 
mechanisms.  Other  species  of  this  genus  present  in  England  are  character- 
ized by  heterostyly  and  incompatibility  devices  to  insure  the  occurrence  of 
outcrossing,  apparently  necessary  to  meet  the  requirements  of  varied  eco- 
logical conditions.  Primula  scofica,  living  in  a  very  specialized  ecological 
niche,  shows  that  such  a  mechanism  has  broken  down.  In  fact,  it  looks  as  if 
the  requirements  of  a  constant  environment  are  met  better  by  populations 
genetically  less  diversified. 


HETEROSIS  IN   POPULATION  GENETICS  153 

Among  animals,  the  largest  ])art  of  which  are  not  sessile  and  therefore  not 
bound  to  the  ground,  the  differentiation  into  two  sexes  offers  the  best  solution 
to  the  problem  of  insuring  a  wide  range  of  crossing  among  different  geno- 
types. But  even  here  we  see  that  special  behavior  patterns  have  been  de- 
veloped for  this  purpose.  These  may  be  courtship  relationships,  sexual  selec- 
tion, dominance  relationships  among  a  group  of  animals,  or  protandry 
mechanisms,  where  the  presence  of  two  sexes  in  hermaphrodites  could  reduce 
the  amount  of  outcrossing  and  therefore  endanger  the  survival  of  the  species. 
Even  among  parthenogenetic  animals,  such  as  Cladoceran  Crustacea,  the  ap- 
pearance of  sexual  generations  after  a  long  succession  of  asexual  ones  seems 
to  depend  upon  extreme  environmental  conditions.  For  its  survival,  the 
species  must  shift  over  to  sexual  reproduction  in  order  to  obtain  a  wider 
range  of  genetic  combinations,  some  of  which  might  be  able  to  survive  under 
the  new  set  of  conditions. 

At  the  level  of  the  chromosome  mechanisms,  several  examples  of  perma- 
nent hybrids  are  known  well  enough  to  be  sure  that  they  play  an  im- 
portant role  for  the  survival  of  some  flowering  plants.  In  animals,  too,  some 
similar  mechanism  may  be  present.  In  a  European  species  of  Drosophila 
which  we  are  studying  now^,  Drosophila  subobscura,  one  finds  that  practically 
every  individual  found  in  nature  is  heterozygous  for  one  or  more  inversions. 
It  looks  as  if  the  species  were  a  permanent  hybrid. 

Rarely,  though,  one  finds  individuals  giving  progeny  wath  homozygous 
gene  arrangement.  Such  cases  have  been  observed  only  three  times:  once 
in  Sweden,  once  in  Switzerland,  once  in  Italy;  and  they  are  very  peculiar 
in  one  respect.  The  three  homozygous  gene  arrangements  are  the  same,  even 
though  the  ecological  and  climatic  conditions  of  the  three  original  popula- 
tions were  as  different  as  they  could  be.  It  looks  as  if  the  species  could 
originate  only  one  gene  arrangement  viable  in  homozygous  condition,  and 
that  this  may  occur  sporadically  throughout  its  vast  distribution  range 
(Buzzati-Traverso,  unpublished). 

At  this  level  too  is  the  fine  example  of  heterozygous  inversions  from  the 
classical  studies  of  Dobzhansky  (1943-1947).  They  have  demonstrated  that 
wild  populations  of  Drosophila  pseudoobscura  show  different  frequencies  of 
inversions  at  different  altitudes  or  in  the  same  locality  at  different  times  of 
the  year.  Variation  in  the  frequency  of  inversions  could  be  reproduced  ex- 
perimentally in  population  cages  by  varying  environmental  factors  such  as 
temperature.  It  is  shown  in  such  a  case  that  natural  selection  controls  the  in- 
crease or  decrease  of  inversions  determining  an  interesting  tyj)e  of  balanced 
polymorphism.  Finally,  according  to  the  investigations  of  Mather  (1942- 
1943)  on  the  mechanism  of  polygenic  inheritance,  it  appears  that  linkage  rela- 
tionships within  one  chromosome,  even  in  the  absence  of  heterozygous  inver- 
sions, tend  to  maintain  a  balance  of  plus  and  minus  loci  controlling  quantita- 
tive characters. 


154  ADRIANO  A.  BUZZATI-TRAVERSO 

We  come  then  to  the  third  level,  that  of  gene  effects.  Here  it  is  well  known 
that  heterozygotes  for  a  certain  locus  sometimes  show  a  higher  viability  or  a 
better  adaptation  to  the  environment  than  either  homozygote.  The  most 
extreme  examples  are  those  of  the  widespread  occurrence  of  lethals  in  wild 
populations  of  Drosophila,  noted  in  the  next  section. 

Every  population  of  plants  and  animals  that  has  been  studied  from 
the  genetic  viewpoint  has  proved  to  be  heterozygous  for  several  loci.  We 
have  now  at  our  disposal  a  large  series  of  data  showing  that  the  phenomenon 
of  genetic  polymorphism  is  frequent  in  plants,  animals,  and  man.  These  offer 
to  the  student  of  evolutionary  mechanisms  the  best  opportunities  to  test  his 
hypotheses  concerning  the  relative  importance  of  selection,  mutation  pres- 
sure, migration,  and  genetic  drift  as  factors  of  evolution.  Wherever  we  find 
a  well  established  example  of  balanced  polymorphism,  such  as  that  of  blood 
groups  and  taste  sensitivity  in  man,  it  seems  safe  to  assume  that  this  is  due  to 
selection  in  favor  of  the  heterozygote.  How  this  selection  actually  may  pro- 
duce an  increase  in  the  chances  of  survival  of  the  heterozygote,  as  compared 
to  both  homozygotes,  is  an  open  question.  When  the  characters  favored  by 
natural  or  artificial  selection  are  the  result  of  several  genes  in  heterozygous 
condition,  the  analysis  becomes  very  difficult  indeed,  as  the  experience  of 
plant  and  animal  breeders  clearly  shows. 

EXPERIMENTS  WITH  HETEROSIS 

The  importance  of  the  problem  of  heterosis  forpopulation-genetical 
studies  is  clearly  shown,  not  only  by  such  general  considerations  and  by  the 
few  examples  mentioned,  but  also  by  the  everyday  experience  of  people 
interested  in  such  lines  of  work.  I  have  come  across  problems  involving 
heterosis  several  times  and  shall  describe  some  results  we  have  obtained 
which  may  be  of  interest  for  the  problem  under  discussion,  especially  at  the 
level  of  single  gene  differences. 

Several  Drosophila  workers  have  been  able  to  show  the  occurrence  of 
heterosis  in  the  fruit  flies.  L'Heritier  and  Teissier  (1933),  Kalmus  (1945),  and 
Teissier  (1947a,  b)  have  shown  that  some  visible  recessive  mutants  of  Dro- 
sophila melanogaster  such  as  ehony  and  sepia  have  a  higher  selective  value  in 
heterozygous  condition  than  either  of  the  corresponding  homozygotes  under 
laboratory  conditions.  Dobzhansky  and  collaborators  in  Drosophila  pseudo- 
obscura,  Plough,  Ives,  and  Child,  as  well  as  other  American  and  Russian 
workers  in  Drosophila  melanogaster,  have  shown  that  recessive  lethals  are 
widely  spread  in  natural  populations.  It  is  generally  accepted  that  such  genes 
are  being  maintained  in  the  population  because  the  heterozygotes  are  being 
selected.  Teissier  (1942,  1944)  has  brought  similar  evidence  under  labora- 
tory conditions  for  Drosophila  melanogaster. 

It  has  been  shown  in  several  populations  of  species  of  the  genus  Drosophila 
that  heterozygous  inversions  are  being  selected,  under  natural  and  ex- 


HETEROSIS  IN  POPULATION  GENETICS  155 

perimental  conditions.  It  seems,  however,  that  the  study  of  selection  in 
favor  of  the  heterozygote  for  single  loci  deserves  more  careful  analysis.  The 
whole  problem  of  heterosis  for  several  genes  affecting  quantitative  characters 
will  be  solved,  I  think,  only  when  the  more  simple  cases  of  heterosis  where 
single  gene  differences  are  involved  shall  be  cleared  up.  I  have  been  lucky 
enough  to  come  across  some  useful  experimental  material  for  the  purpose. 

For  a  number  of  years  I  have  kept  about  one  hundred  different  wild 
stocks  of  Drosophila  melanogaster  coming  from  diflferent  geographical  locali- 
ties. Such  stocks  were  maintained  by  the  usual  Drosophila  technique  of 
transferring  about  once  a  month  some  30-40  flies  from  one  old  vial  to  a  new 
one  with  fresh  food.  About  twice  a  year  I  look  at  the  flies  under  the  micro- 
scope. Since  all  such  stocks  were  wild  type,  no  change  by  contamination  was 
expected,  as  these  stocks  were  phenotypically  alike.  Contamination  by  mu- 
tants kept  in  the  laboratory  could  not  have  produced  any  appreciable  result, 
owing  to  the  well  known  fact  that  both  under  laboratory  and  natural  condi- 
tions mutants  are  generally  less  viable  than  the  normal  type.  To  my  sur- 
prise, however,  I  happened  to  observe  at  two  different  times,  in  two  different 
wild  stocks,  that  a  fairly  large  number  of  the  flies  showed  an  eye  color  much 
lighter  than  the  normal.  These  two  mutants  proved  to  be  indistinguishable 
recessive  alleles  at  the  same  locus  in  the  third  chromosome.  The  presence 
of  the  homozygotes  has  been  checked  at  different  times  over  a  number  of 
years. 

In  the  summer  of  1947  while  collecting  flies  in  the  wild  for  other  purposes, 
I  found  in  the  neighborhood  of  Suna,  a  small  village  on  the  western  shore  of 
the  Lake  Maggiore,  in  Northern  Italy,  several  individuals  of  both  sexes  show- 
ing the  same  eye  color.  From  these  a  homozygous  stock  for  such  mutant  was 
obtained.  Crossing  tests  proved  that  it  was  another  allele  of  the  same  locus  as 
the  above  mentioned.  The  occurrence  of  several  individuals  mutant  for  an 
autosomal  recessive  within  a  free  living  population  was  remarkable  enough. 
But  finding  that  the  same  gene  was  concerned  as  in  the  laboratory  stocks,  I 
suspected  that  such  a  mutant  might  have  a  positive  selective  value,  both 
under  laboratory  and  natural  conditions. 

I  began  an  experiment  to  check  this  point.  Two  populations  in  numerical 
equilibrium  were  started,  applying  the  method  previously  used  by  Pearl  for 
the  study  of  population  dynamics  of  Drosophila,  described  in  detail  else- 
where (1947a).  Sixteen  light-eyed  individuals,  eight  males  and  eight  females, 
were  put  together  in  one  vial  with  sixteen  wild  type  flies.  The  gene  frequency 
at  the  beginning  of  the  experiment  was  therefore  .5.  Under  the  experimental 
conditions  the  population  reached  an  equilibrium  in  respect  to  the  amount 
of  available  food  at  about  700-900  flies  per  vial.  After  about  twenty  genera- 
tions, assuming  that  each  generation  takes  15  days,  the  frequency  of  recessive 
homozygotes  was  about  40  per  cent.  Assuming  random  mating  within  the 
population,  taking  the  square  root  of  .40  one  gets  a  gene  frequency  for  the 


156  ADRIANO  A.  BUZZATI-TRAVERSO 

light-eyed  gene  of  about  .63.  Since  in  both  parallel  populations  the  gene  fre- 
quency was  similar,  one  could  conclude  that  selection  had  favored  the  mutant 
type,  shifting  its  frequency  from  .5  to  .63  in  the  course  of  about  twenty 
generations. 

Such  an  experiment  did  prove  that  the  mutant  gene  had  a  positive  selec- 
tive value.  It  was  impossible  to  know  whether  in  the  long  run  it  would  have 
eventually  eliminated  its  normal  allele  from  the  population.  At  this  stage,  I 

0.875^ 


Fig.  9.1 — Variation  in  the  frequency  of  the  Hght-eyed  gene  in  selection  experiments.  In  the 

abscissae  is  the  number  of  generations,  in  the  ordinates  the  gene  frequency.  Each  line 

represents  a  single  experiment  on  an  artificial  population. 

have  begun  a  new  experiment  along  the  same  lines,  but  with  different  gene 
frequencies  to  start  with.  Two  populations  were  started  with  2  males  and  2 
females  of  the  mutant  type,  plus  14  males  and  14  females  of  the  normal  type. 
Two  populations  were  started  with  16  mutant  and  16  wild  flies,  and  two 
populations  with  28  mutant  and  4  wild  type  flies. 

I  had,  therefore,  at  the  beginning  of  the  experiment  six  populations.  Two 
had  a  gene  frequency  of  the  light-eyed  mutant  approximately  equal  to  .125. 
Two  had  a  gene  frequency  of  .5,  and  two  had  a  gene  frequency  of  .875. 
Figure  9.1  shows  the  result  of  such  an  experiment  after  about  fifteen  genera- 
tions. Crossings  of  wild  type  males,  taken  from  the  populations,  with  homo- 
zygous recessive  females  showed  that  there  was  no  significant  departure  from 
random  mating  within  the  population.  The  gene  frequencies  indicated  on  the 


HETEROSIS  IN  POPULATION  GENETICS 


157 


orclinates  were  obtained  by  taking  the  square  root  of  the  observed  frequencies 
of  homozygous  recessives. 

The  following  conclusions  can  be  drawn:  (1)  the  three  experimental  popu- 
lations, each  being  run  in  duplicate,  have  reached  the  same  gene  frequency 
at  about  the  .579  point;  (2)  natural  selection  has  been  acting  on  the  three 
populations  producing  the  same  end  results,  irrespective  of  the  initial  gene 
frequency;  (3)  natural  selection  has  been  acting  in  favor  of  the  heterozygous 
flies;  and  (4)  the  homozygous  mutant  seems  to  be  slightly  superior  in  its 
survival  value  to  the  homozygous  normal  allele. 

It  was  of  considerable  interest  to  determine  whether  the  intensity  of  selec- 
tion operating  in  the  three  experiments  was  the  same.  Since  the  three  experi- 
mental curves  (each  being  the  mean  of  the  two  duplicate  populations)  could 
not  be  compared  directly.  Dr.  L.  L.  Cavalli  elaborated  a  mathematical 
analysis  of  the  problem  (Cavalli,  1950).  The  function  of  gene  frequency  linear 
with  time  F,  when  the  heterozygote  is  at  an  advantage,  is  given  by: 

Y  =   Qe^Og  p-\-pe\0g    q  -\0g[pe-p]  , 

where  p  and  q  are  the  gene  frequencies  at  the  beginning  of  the  experiment  in 
a  random  breeding  population,  and  pe  and  qe  are  the  equilibrium  frequencies. 
By  means  of  this  function  it  is  possible  to  transform  the  experimental  curves 
to  linear  ones.  Results  can  then  be  plotted  graphically  for  the  three  experi- 
ments. Fitting  straight  lines  with  the  method  of  maximum  likelihood,  one 
obtains  the  following  values  for  the  constants  of  the  linear  regression  equa- 
tion: 


Experi- 
ment 

Initial  Gene 
Frequency 
(Observed) 

Slope 

Position 

Initial  Gene 

Frequency 

(Theoretical) 

1 

2 

3 

.500 
.125 
.875 

.0879 
.0631 
.0726 

+  1.21 
-    .41 

+   .27 

.425 
.100 
.830 

The  position  is  the  transformed  value  of  the  initial  gene  frequency  which 
is  given  in  the  last  column,  and  is  in  good  agreement  with  the  experimental 
value.  If  one  tests  the  parallelism  of  the  three  regression  lines  so  obtained,  one 
gets  a  chi  square  of  4.0  with  two  degrees  of  freedom.  Parallelism  therefore 
seems  to  be  satisfactory.  This  implies  that  the  intensity  of  selection  is  inde- 
pendent of  initial  conditions. 

If  we  take  these  results  together  with  the  two  independent  occurrences  of 
the  same  mutant  gene  in  different  genotypical  milieus,  it  seems  safe  to  main- 
tain that  such  a  gene  has  a  positive  selective  value  with  respect  to  its  normal 
allele,  and  that  selection  is  acting  mainly  through  a  typical  heterosis  mecha- 
nism. It  is  worth  while  to  stress  that  this  gene  was  found  both  in  natural  and 


158  ADRIANO  A.  BUZZATI-TRAVERSO 

experimental  conditions.  The  exceptional  occurrence  of  many  mutant  indi- 
viduals in  a  free  living  population  can  be  accounted  for  by  assuming  that 
they  have  a  higher  selective  value. 

BASIS  FOR  SUPERIORITY  OF  THE  HETEROZYGOTE 

It  would  be  interesting  to  try  to  find  out  how  selection  discriminates 
against  both  normal  and  mutant  homozygotes.  I  am  just  beginning  to  attack 
this  problem. 

Dr.  E.  Caspar!  has  some  interesting  results  on  a  similar  problem,  and  I 
wish  to  thank  him  for  permission  to  quote  them  (Caspari,  1950).  In  free 
living  populations  of  the  moth  Ephestia  kuhniella,  this  author  has  observed 
a  balanced  polymorphism,  whereby  individuals  having  brown  colored  and 
red  colored  testes  occur  in  various  numbers.  The  character  brown  behaves  as 
a  complete  dominant  with  respect  to  red.  The  polymorphism  seems  to  be 
determined  by  a  higher  selective  value  of  the  heterozygote.  It  has  been  pos- 
sible to  show  that  the  heterozygote  is  equal  or  superior  to  the  homozygous 
recessive  and  the  latter  is  superior  to  the  homozygous  dominant  with  respect 
to  viability.  It  was  found  that,  while  the  heterozygote  is  equal  or  superior  to 
the  homozygous  dominant,  the  homozygous  brown  is  superior  to  the  homozy- 
gous red  with  respect  to  mating  activity.  The  dominance  relationships  of  such 
two  physiological  characters  are  therefore  reversed. 

There  is  no  decisive  evidence  for  heterosis  for  any  of  the  characters 
studied.  The  recessive  for  the  testis  color  acts  as  dominant  with  respect  to 
viability,  and  the  dominant  testis  color  acts  as  dominant  with  respect  to 
mating  behavior.  The  net  result  is  a  selective  advantage  of  the  heterozygote 
that  can  account  for  the  observed  polymorphism.  This  seems  a  good  ex- 
ample of  how  a  heterosis  mechanism  can  be  determined  by  the  behavior  of 
two  visible  alleles  in  heterozygous  condition.  It  is  hoped  that  similar  analyses 
will  be  developed  for  other  cases  of  balanced  polymorphism. 

The  search  for  clear-cut  examples  of  heterosis  depending  on  single  genes 
seems  to  me  the  most  promising  line  of  attack  on  the  general  problem  under 
discussion.  If  I  could  find  another  gene  behaving  in  a  way  similar  to  the  one 
I  have  studied  in  Drosophila  melanogaster,  and  could  study  the  interaction  of 
the  two,  it  would  be  possible  to  go  a  step  further  in  the  analysis  of  heterosis 
mechanisms.  The  evidence  derived  from  such  single  genes  being  favored  in 
heterozygous  condition  is  likely  to  be  very  useful  in  more  complex  condi- 
tions where  the  action  of  several  genes  is  involved. 

When  we  come  to  consider  the  selective  advantage  of  polygenic  charac- 
ters, even  in  such  an  easy  experimental  object  as  Drosophila,  the  problem 
becomes  very  entangled  indeed.  In  recent  years  I  have  been  studying,  for 
example,  a  number  of  quantitative  characters  being  favored  by  natural 
selection  in  artificial  populations  in  numerical  equilibrium,  such  as  the  ones 
I  have  been  speaking  about.  I  have  set  in  competition  at  the  beginning  of  one 


HETEROSIS  IN  POPULATION  GENETICS  159 

experiment  two  stocks  differing  for  visible  mutants.  One  stock  was  while- 
and  Bar-eyed,  the  other  stock  was  normal  for  both  characters.  The  two 
stocks  differed,  too,  in  a  number  of  quantitative  characters  such  as  fecundity, 
fertility,  rate  of  development,  longevity,  and  size. 

After  about  thirty  generations  the  two  mutant  genes  had  been  wiped  out. 
This  could  have  been  expected  on  the  basis  of  previous  data  of  L'Heritier  and 
Teissier  on  the  elimination  of  such  genes  in  artificial  populations.  At  that 
time,  however,  I  did  not  discard  the  populations,  but  kept  them  going  for 
some  seventy  more  generations.  All  the  individuals  present  in  the  popula- 
tions were  phenotypically  normal.  But  testing  from  time  to  time  the  values 
of  the  above  mentioned  characters,  I  could  establish  that  natural  selection 
was  continuously  operating  and  favoring  higher  fecundity,  higher  fertility, 
higher  longevity,  and  quicker  developmental  rate  throughout  the  four  years 
that  the  experiment  lasted.  At  the  end,  the  flies  present  in  the  population 
were  superior  by  a  factor  of  more  than  six  to  the  mean  of  the  considered 
characters  in  the  two  original  parental  stocks.  When  I  measured  such  values 
in  the  Fi  hybrids  between  the  two  stocks  I  could  observe  values  higher  than 
those  obtained  after  more  than  one  hundred  generations  of  selection. 

The  selection  experiment  could  then  be  interpreted  in  two  different  ways. 
Either  (a)  selection  had  picked  up  a  new  genotype  made  out  of  a  new  com- 
bination of  polygenes  derived  from  the  two  parental  stocks,  or  (b)  selection 
had  just  preserved  by  means  of  a  heterosis  mechanism  a  certain  amount  of 
heterozygosity,  which  was  at  its  highest  value  at  the  beginning  of  the  experi- 
ment. The  fact  that  in  the  course  of  the  experiment  the  factors  had  been 
steadily  improving  seemed  to  be  against  hypothesis  b,  but  I  could  not  be  sure 
that  was  the  case. 

I  then  set  up  a  new  selection  experiment,  whereby  I  put  in  competition  the 
original  stock  white  Bar  with  the  normal  type  derived  from  the  population 
which  had  been  subjected  to  natural  selection  for  more  than  one  hundred 
generations.  The  result  was  clear.  The  genes  white  and  Bar  were  elimi- 
nated in  this  second  experiment  at  a  much  higher  rate  than  in  the  first  ex- 
periment. In  the  first  experiment  the  gene  frequency  of  the  gene  Bar  after 
ten  generations  had  dropped  from  .50  to  .15.  In  the  second  experiment,  after 
as  many  generations,  the  Bar  gene  frequency  had  dropped  from  .50  to  .03. 
It  seems  that  the  genotype  produced  by  a  hundred  generations  of  natural 
selection  under  constant  conditions  was  so  much  better  adapted  to  its  en- 
vironment that  it  could  get  rid  of  the  competing  genes  with  much  greater 
ease  than  the  original  wild  type  flies.  But  could  it  not  be  that  all  or  at  least 
part  of  this  result  could  be  accounted  for  by  the  action  of  some  heterosis 
effect? 

Another  example  of  a  similarly  puzzling  condition  is  an  experiment  on 
artificial  populations  under  way  now  in  my  laboratory.  I  would  like  to  find 
out  whether  it  is  possible  to  produce  so-called  small  mutations  or  polygenic 


160  ADRIANO  A.  BUZZATI-TRAVERSO 

mutations  with  X-rays,  and  whether  an  increase  in  the  mutation  rate  may 
speed  up  the  evolutionary  rate  under  selection  pressure. 

For  this  purpose  I  have  set  up  four  artificial  populations  starting  from  an 
isogenic  stock  of  Drosophila  melanogaster.  One  of  these  is  being  kept  as  con- 
trol while  the  other  three  get,  every  two  weeks,  500,  1000,  and  2000  r-units 
respectively.  At  the  start,  and  at  various  intervals,  I  am  measuring  fecun- 
dity, fertility,  and  longevity  of  the  flies.  The  few  data  so  far  collected  show 
clearly  that  in  the  irradiated  populations  the  percentage  of  eggs  that  do  not 
develop  is  much  higher  than  in  the  control.  This  is  due  to  the  effect  of 
dominant  and  recessive  lethals.  But  the  startling  result  is  that  the  fecundity, 
measured  by  the  number  of  eggs  laid  per  day  by  single  females  of  the  irradi- 
ated populations,  is  higher  than  in  the  control  series.  Probably  X-rays  have 
produced  a  number  of  mutations  for  higher  fecundity  which  have  been  ac- 
cumulated by  natural  selection  in  the  course  of  the  experiment.  But,  are  spe- 
cific mutations  for  higher  fecundity  being  produced,  or  am  I  dealing  with 
heterosis  phenomena  dependent  upon  nonspecific  mutants? 

These  few  examples  from  my  own  experience  with  population-genet ical 
studies  show,  I  think,  how  important  the  heterosis  phenomenon  can  be  in  our 
field  of  work.  Both  in  natural  and  artificial  populations,  heterosis  seems  to 
be  at  work,  making  our  analysis  rather  difficult,  but  stimulating  as  well. 
Closer  contacts  between  students  of  selection  and  heterosis  in  plant  and 
animal  breeding  and  students  of  evolutionary  problems  are  to  be  wished. 
Let  us  hope  that  a  higher  level  of  hybridization  between  various  lines  of 
investigation  might  become  permanent,  since  it  surely  will  make  our  studies 
more  vigorous  and  better  adapted  to  the  requirements  of  a  rapidly  growing 
science. 


HAROLD  H.  SMITH 

Cornell  Universify 


Chapter   10 

Fixing  Tronsgressive  Vigor 
in  Nicofiona  Rustica 


Hybrid  vigor  has  been  observed  to  varying  degrees  among  certain  inter- 
varietal  hybrids  of  the  self-pollinated  cultivated  species  Nicotiana  rustica  L. 
(Bolsunow,  1944;  East,  1921).  In  experiments  undertaken  to  obtain  a  larger 
;V.  rustica  plant  giving  increased  yield  of  nicotine,  it  was  reported  (Smith 
and  Bacon,  1941)  that  inbred  lines  derived  as  selections  from  hybrids  among 
three  varieties  exceeded  the  parents  and  Fi's  in  plant  height,  number  of 
leaves,  or  size  of  the  largest  leaf. 

The  general  experience  of  breeders  of  self-pollinated  plants  has  been  that 
improved  varieties  can  be  developed  through  hybridization  followed  by  selec- 
tion and  inbreeding,  to  fix  desirable  transgressive  characteristics.  Yet  it  is 
difficult  to  find  data  from  which  quantitative  relationships  of  parents,  Fi,  and 
transgressive  inbred  can  be  adequately  evaluated;  as  from  replicated  and 
randomized  experiments  in  which  the  generations  have  been  grown  at  the 
same  time  under  comparable  conditions.  In  view  of  the  increasing  number 
of  reports  on  hybrid  vigor  in  self-pollinated  crop  plants  and  its  suggested 
utilization  (Ash ton,  1946),  it  was  considered  opportune  to  present  relevant 
data  accumulated  on  N.  rustica. 

Since  methods  of  partitioning  phenotypic  variance  have  become  generally 
available  there  was  additional  interest  in  making  further  study  of  the  iV. 
rustica  material.  Breeding  results  obtained  in  advanced  selections  could  be 
related  to  the  heritability  estimated  from  data  on  early  generations. 

MATERIALS  AND  METHODS 

Four  varieties  of  Nicotiana  rustica  were  used  in  these  experiments.  Three 

of  them — hrasilia  strain  34753,  Olson  68,  and  tall  type  have  been  described 

*  Published  as  Paper  No.  261 ,  Department  of  Plant  Breeding,  Cornell  University,  Ithaca 
New  York. 

161 


162  HAROLD  H.  SMITH 

in  Smith  and  Bacon  (1941).  The  fourth  was  received  originally  from  the 
director  of  the  Tabak-Forschungsinstitut,  Baden,  Germany,  under  the  name 
of  texana,  a  designation  which  we  have  retained.  It  is  a  small,  early-maturing 
type.  The  four  parental  varieties  were  of  highly  inbred  stocks  maintained  by 
the  Division  of  Tobacco,  Medicinal  and  Special  Crops  of  the  United  States 
Department  of  Agriculture.  The  earlier  part  of  the  breeding  program  was 
carried  out  while  the  writer  was  associated  with  this  organization. 

The  advanced  selection,  designated  Al,  used  in  these  experiments  has  a 
complex  genetic  history  of  crossing,  backcrossing,  and  inbreeding.  This  can 
be  briefly  summarized  by  stating  that  its  ultimate  composition  was,  on  an 
average,  60  per  cent  34753,  22  per  cent  Olson,  12  per  cent  tall  type,  and  6 
per  cent  texana.  About  82  per  cent  of  the  Al  genotype  was,  on  chance  alone, 
contributed  by  the  two  most  vigorous  parents,  34753  and  Olson  68.  This 
calculation  does  not  take  into  account  any  differential  effect  of  selection  on 
changing  the  frequency  of  genes  introduced  from  diverse  parental  origins. 
Observation  of  the  Al  phenotype  led  us  to  believe  that  selection  had  further 
increased  the  proportion  of  genes  from  the  two  most  vigorous  parents. 

In  1947  the  four  parents,  the  six  possible  F/s,  the  three  double  crosses, 
and  the  F4  generation  (preceded  by  three  generations  of  inbreeding)  of  line 
Al  were  grown  in  a  randomized  complete  block  design  with  fifteen  plants  in 
each  plot  and  replicated  six  times.  In  1949  the  two  most  vigorous  varieties 
(Olson  68  and  34753),  the  Fi,  F2,  backcrosses  of  the  Fi  to  each  of  its  parents, 
and  the  Fe  generation  of  line  Al  were  grown  in  a  randomized  complete  block 
design  with  twenty  plants  in  each  plot  and  replicated  eight  times. 

Measurements  were  made  on  plant  height,  number  of  leaves  or  nodes,  and 
length  of  the  largest  leaf.  In  addition,  data  were  taken  on  the  width  of  the 
largest  leaf,  number  of  days  from  planting  to  appearance  of  the  first  flower, 
and  total  green  weight  of  individual  plants. 

Typical  plants  of  Olson  68,  34753,  the  Fi  between  these  two  varieties,  and 
selection  Al  are  illustrated  in  Figure  10.1. 

EXPERIMENTAL  RESULTS 

Data  obtained  from  the  1947  and  1949  plantings  are  summarized  in 
Tables  10.1  and  10.2,  respectively. 

Phenotype-Genotype    Relations 

Preceding  further  biometrical  analysis  of  the  data,  tests  for  evidence  of 
differential  environmental  effects  and  genetic  interactions  were  made.  For 
the  former,  the  relation  between  genotype  mean  and  non-heritable  variabil- 
ity was  determined  by  comparing  means  and  variances  of  the  parents  and  Fj 
(1949  data.  Table  10.2).  For  the  characters  plant  height  and  leaf  length,  the 
variances  were  unrelated  to  the  means  and  the  parental  variances  were  not 
significantly  different^from  each  other.  For  node  number,  however,  the 


Fig.  10.1 — Typical  field-grown  (1949)  ]ilants  of  Nicotiana  riistica.  Left  to  right:  Olson  68, 
brasilia  strain  34753,  Fi  Olson  68  X  34753,  and  selection  A^Fe).  The  scale  shown  at  the 

left  is  in  inches. 


TABLE  10.1 

PLANT  CHARACTERISTICS  IN  PARENTAL  VARIETIES,  HYBRIDS, 
AND  AN  INBRED  SELECTION  OF  NICOTIANA  RUSTICA* 


Type 

Plant 
Height 

Leaf 
Number 

Leaf 
Length 

Mean 
Leaf 
Width 

Mean 
Days 

TO 

Ma- 
ture 

Gen- 
era- 
tion 

Mean 

Total 

within 
Plot 

Mean 

Total 

within 

Plot 

Mean 

Total 

within 

Plot 

Mean 
Green 

Wgt. 

Per 
Plant 

d.f. 

Var. 

d.f. 

Var. 

d.f. 

Var. 

Pi 

Olson  68  (/4 ) 

in. 

49.9 

29.0 

46.7 

33.6 

73 
84 
83 
83 

17.8 
20.0 
30.3 
11.1 

18  8 
15.8 
16  0 
12.7 

72 
82 
83 
83 

4.68 
7.96 
1.36 
1.04 

in. 

11.7 
8.7 
6  2 
6  6 

72 
82 
83 
83 

1.07 
1.43 
1.24 
1.01 

in. 

8.7 
7.5 
5.5 
5.4 

70.1 
66.6 
48.0 
40.9 

lbs. 
1   51 

34753  (B) 

0  89 

Tall  type  (C)          

0  SI 

Texana (D) 

0  48 

Average 

01sonX34753UX5).  . 

OlsonXtall(^XC) 

Olson Xtexana  {A  XD) . 

34753  Xtall(BXC) 

34753Xtexana(SX£>). 
Tall  Xtexana  (CXC)  .  . 

Average 

iAXB)X(CXD) 

{AXC)X(BXD} 

{AXD)XiBXC) 

Average 

39.8 

19.8 

15.8 

3.76 

8.3 

1.19 

6.8 

56.4 

0.85 

Fi 

48.5 
42.9 
40.1 
47.1 
40.3 
44.2 

74 
80 
81 
84 
83 
84 

58.3 
25.4 
20  8 
45.2 
28.1 
29.3 

16.8 
13  0 
11.2 
16  6 

14.4 
15.5 

74 
80 
81 
84 
83 
83 

7.18 
7.74 
4  99 
1.77 
6  34 
1.49 

10  8 

10.1 

10.6 

7.8 

8  7 

7.7 

74 
80 
81 
84 
83 
83 

1.92 
1.82 
1.49 
1.37 
2.30 
1.88 

8.7 
9.6 
9.7 
6.7 
7.4 
7.0 

75.0 
65.1 
70.6 
50.5 
60.0 
51.8 

1.47 
1.16 
1.13 
0.76 
0.93 
0.95 

43.8 

34.5 

14.6 

4.92 

9.3 

1.80 

8.2 

62.2 

1.07 

FiXFi 

41.9 
39  6 

42.5 

75 
82 
81 

25.1 
61.1 
39.9 

14.0 
12.4 
13.9 

74 
79 
80 

10.60 

9.29 

10  80 

8  8 

9  8 
9.7 

74 
79 
80 

2.36 
3.11 
2.50 

7.7 
8.8 
8.7 

61.9 
66.0 
60.4 

0.86 
0.99 
1.06 

41.3 

42.0 

13.4 

10.23 

9.4 

2.66 

8.4 

62.8 

0  97 

Selection  Al 

F4 

54.9 

78 

78.8 

19.9 

77 

S.28 

10.4 

77 

1.51 

8.0 

79.6 

1  83 

Least  si 
5%le 
l%le 

?nificant  diff.  at 

vel 

2.68 
3.56 

1.22 
1.62 

0.89 
1.19 

0.76 
1.00 

4.11 
5.46 

0  25 

vel 

0  34 

*  Summary  of  1947  data. 


164 


HAROLD  H.  SMITH 


means  and  non-heritable  variances  were  linearly  related  for  both  1947  and 
1949  data,  and  the  parental  variances  were  significantly  different. 

Tests  to  reveal  the  presence  or  absence  of  non-allelic  interactions  were  then 
made  according  to  the  method  proposed  by  Mather  (1949).  Results  are 
shown  in  Table  10.3.  No  significant  deviations  from  zero  were  found  if  the 
level  of  significance  was  taken  as  P  ^  .01.  In  each  test,  however,  the  P  values 
for  number  of  nodes  were  less  than  for  plant  height  or  leaf  length,  possibly 
owing  to  non-additive  gene  effects. 

It  was  concluded,  on  the  basis  of  these  tests,  that  for  the  two  characters 


TABLE  10.2 

PLANT  CHARACTERS  IN  THE  TWO  MOST  VIGOROUS  VARIETIES  OF 

N.  RUSTIC  A,  THEIR  F,,  F2,  AND  FIRST  BACKCROSS  PROGENY 

AND  IN  SELECTION  AKFe)* 


Gen- 
era- 
tion 


Pi 
P2 
Fi 
F2 
Bi 
B2 
Fe 


Type 


Olson  68 

34753 

Olson  68X34753... 
(Olson X34753)  self. 

FiXOlson  68 

FiX34753 

Selection  Al 


Plant  Height 


Mean 


in. 
47.8 
28.7 
43.2 
40.6 
47.3 
36.2 
55.6 


Total  with- 
in Plot 


d.f. 


141 
143 
140 
149 
149 
148 
133 


Var. 


15.46 
22.63 
39.18 
99.19 
40.28 
101.50 
29.77 


No.  OF  Nodes 


Mean 


24. 
21. 

22. 
23. 
24. 
21. 


31.0 


Total  with- 
in Plot 


d.f. 


136 
106 
110 
119 
138 
117 
126 


Var. 


3. 
10. 

8. 


45 
10 
60 


10.52 
10.49 


45 
44 


Leaf  Length 


Mean 


in. 
11 
10. 
11 
11 
11 
10.8 
12.0 


Total  with- 
in Plot 


d.f. 


142 
127 
131 
142 
144 
135 
141 


Var. 


0.68 
0.81 
0.63 


08 
10 


0.95 
0.69 


Least  significant  diff.  at 

5%  level 2.55 

1%  level 3.42 


1.37 
1.83 


0.49 
0.66 


*  Summary  of  1949  data. 

TABLE  10.3 
SCALING  TESTS  FOR  AVERAGE  ADDITIVENESS  OF  GENE  EFFECTS* 


Character 


Test  A 


Dev. 


Var. 


Plant  height.  3.6  2.86 
No.  nodes...'  2.8  1.50 
Leaf  length..    0.9  0.47 


Test  B 


Dev. 


Var. 


Test  C 


Dev. 


Var. 


2. 13. 03-. 04  0.54.13  0. 25'. 80-. 81 -0.515.48  0. 13  .89-. 90 
2. 30. 02-. 03-1. 11. 71  0.84. 40-. 41  3.5  6.18  1.40. 16-. 17 
1.32,. 18-. 19,    0.00.47  0.00    1.00   i    0.5   1.77  0.38  .  70-.  71 


*  Based  on  1949  means. 


FIXING  TRANSGRESSIVE  VIGOR  IN  NICOTIANA  RUSTICA  165 

plant  height  and  leaf  length,  the  data,  as  taken,  could  be  used  without 
serious  error  in  partitioning  the  variance  of  segregating  generations.  For 
node  number  it  was  indicated  that  some  correction  should  be  made  with 
the  data  before  further  biometrical  analysis  was  undertaken. 

Mather  suggested  that  difficulties  of  the  sort  encountered  in  these  data 
with  node  number  may  be  overcome  by  finding  a  transformation  of  scale  on 
which  they  would  be  minimized.  The  transformations  \/X,  X'^,  X^,  and 
\/a  -\-  bx  on  the  individual  measurements  were  made.  In  the  latter  transfor- 
mation b  is  the  linear  regression  coefficient  and  a  the  intercept.  Also,  for 
^ya  +  bx  —  K,  \/—K  was  taken  as  —\/K.  In  some  cases  the  transforma- 
tions reduced  the  departure  from  the  preferred  relationship  in  one  test,  only 
to  make  the  transformed  data  less  preferable  by  another  test.  No  transforma- 
tion tried  resulted  in  a  consistent  improvement  over  the  original  scale,  and 
consequently  none  was  used. 

It  is  evident  that  the  significantly  different  variances  in  node  number  of 
the  two  parental  types  were  due  mainly  to  different  interactions  between 
genotype  and  environment.  From  previous  experience  we  know  that  under 
ideal  conditions  of  growth,  Olson  68  and  strain  34753  show  approximately  the 
same  variability.  The  adverse  weather  conditions  of  the  1949  season  were  ob- 
served to  have  a  more  deleterious  effect  on  leaf  number  in  strain  34753.  Con- 
sequently it  was  considered  that  the  greater  variability  of  this  variety,  com- 
pared to  Olson  68,  could  be  attributed  to  a  greater  phenotypic  interaction 
between  genotype  and  environment.  In  view  of  these  relationships,  the 
analysis  of  the  data  on  node  number  was  approached  in  another  way,  as 
mentioned  below  under  "Partitioning  Phenotypic  Variance." 

First  Generation   Hybrids 

Deviations  of  the  Fi  means  from  mid-parent  values  (arithmetical  average 
between  parental  means)  can  be  used  to  estimate  the  preponderance  of 
dominant  gene  effects,  acting  in  one  direction,  at  loci  by  which  the  parental 
complements  differ.  Mid-parent  values  were  calculated  from  the  1947  da- 
ta on  the  four  original  varieties.  The  results  for  each  line  are  summarized  in 
Table  10.4.  The  data  shown  were  obtained  by  first  calculating  the  difference 
between  the  Fi  mean  and  the  mid-parent  (Fi  —  MP)  for  each  cross,  then  tak- 
ing the  average  of  the  differences  for  each  group  of  three  F/s  involving  the 
parent  variety  under  consideration.  The  ratio  of  the  deviation  of  the  Fi  from 
the  mid-parent  to  half  the  parental  difference,  Fi  —  MP/\{P2  —  Pi) ,  is  a  meas- 
ure of  the  relative  potence  (Mather,  1949;  Wigan,  1944)  of  the  gene  sets.  Po- 
tence  ratios,  calculated  from  averages,  are  shown  in  Table  10.4.  For  plant 
height  and  leaf  length  the  Fi  means  fall,  on  an  average,  about  .6  of  the  dis- 
tance from  the  mid-parent  toward  the  larger  parent.  For  leaf  number  the  Fi 
means  fall,  on  an  average,  about  .7  of  the  distance  from  the  mid-parent 
toward  the  smaller  parent. 


166 


HAROLD  H.  SMITH 


The  Fi's  were  taller  and  had  larger  leaves,  on  an  average,  than  the  mid- 
parent.  It  was  concluded,  therefore,  that  a  preponderance  of  dommant+ 
genes  was  mvolved  in  determining  differences  in  plant  height  and  leaf  length. 
In  the  development  of  the  parent  varieties,  selection  resulted  in  the  accumu- 
lation of  dominant-f  modifiers,  as  is  usually  the  case  in  naturally  cross- 
pollinated  plants. 

The  result  with  the  character  leaf  number  was  different  in  that  the  Fi  had 
fewer  leaves,  on  an  average,  than  the  mid-parent.  Evidently,  in  the  evolution 
of  the  varietal  gene  sets,  there  had  been  accumulated  a  preponderance  of 
recessive+ modifiers  (or  dominant  genes  ior fewer  leaves)  at  the  loci  by  which 

TABLE  10.4 

DIFFERENCE  BETWEEN  THE  Fi  AND  MID-PARENT  (Fi  -  MP)  AND  THE 
POTENCEt  RATIO  IN  INTERVARIETAL  HYBRIDS.  BOTH  V.ALUES  ARE 
EXPRESSED  AS  THE  AVERAGE  FOR  EACH  VARIETY  IN  CROSSES  WITH 
THE  OTHER  THREE  VARIETIES* 


Variety 

PL.4NT  Height 

No.  Leaves 

Leaf  Length 

Fi-MP 

Potencef 

Fi-MP 

Potence 

Fi-MP 

Potence 

Olson  68 

34753    

in. 

+0.7 
+9.1 

+2.6 
+3.8 
+4.0 

+0.10 

+  1.26 
+0.46 
+0.68 
+0.62 

-3.2 
+0.1 
-0.9 
-1.1 
-1.2 

-1.62 
+0.09 
-0.87 
-0.53 
-0.73 

in. 

+0.9 
+0.3 
+0.8 
+  1.1 
+0.8 

+0.43 
+0.33 

Tall              .      .  . 

+0.63 

Texana 

+0.97 

Average .... 

+0.59 

*  1947  data.  _    _    _ 

t  Potence  =  Fi-MP/kiPr-Pi). 

the  parents  differed.  There  can  be  little  doubt  that  selection  for  mafty  leaves 
was  practiced  in  producing  the  parent  types.  This  is  especially  true  for  Olson 
68  which  was  developed  from  hybrid  origin  by  the  late  Mr.  Otto  Olson 
(Smith  and  Bacon,  1941)  by  selection  for  plants  yielding  large  amounts  of 
nicotine.  In  crosses  with  Olson  68,  the  Fi  was  consistently  below  the  mid- 
parent.  This  result,  interpretable  as  due  to  an  accumulation  of  a  preponder- 
ance of  recessive  genes  for  the  character  favored  by  selection,  might  be  ex- 
pected occasionally  in  naturally  self-pollinated  plants.  Dominance  is  of  less 
importance  here  than  in  cross-pollinated  organisms,  since  selection  is  largely 
a  matter  of  sorting  out  superior  homozygous  combinations. 

The  1949  results  (Table  10.2)  on  Olson  68  X  34753  were  consistent  with 
those  of  1947  discussed  above. 

Double  Crosses 

The  three  possible  double  crosses  involving  all  six  Fi  hybrids  of  four  varie- 
ties were  grown  in  1947  in  order  to  obtain  evidence  on  genie  interactions  by 
comparing  experimental  results  with  predicted  values.  The  latter  were  made 


FIXING  TRANSGRESSIVE  VIGOR  IN  NICOTIANA  RUSTICA 


167 


in  the  manner  employed  in  corn  breeding,  namely  Jenkins'  method,  in  which 
the  average  of  the  four  Fi's  not  contributing  to  the  double  cross  was  used. 
These  comparisons  are  shown  in  Table  10.5  for  the  three  plant  characters 
studied.  The  differences  between  observed  means  and  j)redicted  values  in  the 
nine  comparisons  made  were  all  within  the  limits  required  for  odds  of  19: 1. 
It  was  concluded  that  the  double  cross  means  for  plant  height,  number  of 
leaves,  and  leaf  length  in  iV.  rustica  could  be  predicted  with  a  high  degree  of 
precision  by  Jenkins'  method.  The  results  indicated  that  there  were  no 

TABLE  10.5 
COMPARISON  BETWEEN  PREDICTED  AND  OBSERVED  VALUES  FOR  PLANT 
HEIGHT,  NUMBER  OF  LEAVES,  AND  LEAF  LENGTH  IN  THREE  DOU- 
BLE CROSSES  INVOLVING  FOUR  VARIETIES  OF  N.  RUSTICA 


Double  Cross 

Observed 

Predicted 

Difference, 
Obs.-Pred. 

Plant  height  {in.) : 

{AXB)X{CXD) 

{AXC)X{BXD) 

{AXD)X{BXC) 

41.9  +  2.68 
39.6  +  2.68 
42.5±2.68 

42.6+1.34 
45.0  +  1.34 
44.0±1.34 

-0.7+3.00 

-5.4  +  3.00 
-1.5±3.00 

Average 

41.3 

43.8 

-2.5 

No.  leaves: 

{AXB)X{CXD) 

{AXC)X{BXD) 

{AXD)X{BXC) 

14.0+1.22 
12.4±1.22 
13.9+1.22 

13.8  +  0.61 
15.0  +  0.61 

14.9  +  0.61 

+0.2  +  1.36 
-2.6  +  1.36 
-1.0  +  1.36 

Average 

13.4 

14.6 

-1.2 

Leaf  length  {in.) : 

{AXB)X{CXD) 

{AXC)X{BXD) 

{AXD)X{BXC) 

8.8  +  0.89 
9.8  +  0.89 
9.7  +  0.89 

9.3  +  0.44 

9.2  +  0.44 

9.3  +  0.44 

-0.5  +  0.99 
+0.6  +  0.99 
+0.4±0.99 

Average 

9.4 

9.3 

+0.1 

A ,  B,  C,  D  represent  the  parent  varieties  as  shown  in  Table  10.1. 


marked  interactions  between  the  genes  or  gene  sets  from  the  four  varieties 
when  combined  in  a  variety  of  associations. 

To  illustrate  this  point,  let  us  assume  that  each  parent  is  homozygous  for 
a  different  allele  at  each  of  two  independent  loci  so  that  A  =  XX YY,  B  = 
X'X'Fip,  C  =  X^XWW\  and  D  =  X^XWW^.  The  Fi's  represent  six  dif- 
ferent combinations  of  these  alleles.  Each  double  cross  contains  all  four  alleles 
of  each  locus  in  four  particular  combinations.  For  example,  the  i)opulation 
{AXB)XiCX  D)  is  1/4A'X2  _^  \/^xX'  +  \/^X^X-  +  \/AX'XHoi  the  X 
locus  and  1/4FF2  +  1/477^  +  1/47^72  +  \/4:Y^YHor  the  Y  locus.  Sixteen 
different  combinations  of  alleles  at  the  two  loci  are  possible  in  this  double 
cross.  Accurate  prediction  of  the  double  cross  value  on  the  basis  of  only  four 
of  these  combinations,  namely:  Fi's  AXC,  AXD,  BXC,  and  B  X  D, 


168 


HAROLD  H.  SMITH 


indicates  that  the  other  12  possible  combinations  do  not  introduce  any  sig- 
nificant non-additive  effects. 

Another  indication  that  epistatic  effects  were  unimportant  in  the  in- 
heritance of  plant  height,  leaf  number,  and  leaf  length  was  afforded  by  the 
evidence  that  the  means  of  the  double  crosses  did  not  differ  significantly  from 
each  other  (Table  10.1). 

The  average  variance  of  the  double  crosses  was  greater  than  that  of  the 
parents  or  Fi's  (Table  10.1),  as  would  be  expected  from  segregation. 

Partitioning  Phenotypic  Variance,  Heritability,  and 
Number  of  Effective  Factors 

Estimates  of  the  magnitude  of  the  non-heritable  variation  (0%),  in  popu- 
lations involving  Olson  68  and  34753  (1949  data),  were  obtained  by  taking 

TABLE  10.6 

ESTIMATES  OF  COMPONENTS  OF  \\\RIABILITY,  NUMBER  OF  EFFECTIVE 
FACTORS  (Ai),  HERITABILITY,  AND  GAIN  FOR  PLANT  HEIGHT,  LEAF 
LENGTH,  AND  NUMBER  OF  NODES  IN  THE  X.  RUSTICA  CROSS  OLSON 
68  X  BRASILIA,  STRAIN  34753* 


Character 

0 

0 

"l 

A'l 

Herit- 
ability 
Per 
Cent 

Gain 

Plant  height 

Leaf  length 

Node  number.  .  . 

25.76  +  15.3 
0.71+  0.45 
7.38+  4.38 

67.32  +  53.5 
1.04+   1.05 
8.16  +  13.00 

113.20  +  71.3 
0.22+  0.69 
2.20+   8.11 

0.81 
1.38 
0.83 

54.9 
11.2 
12.4 

1.74 
0.91 

2.42 

*  1949  data. 


an  average  of  the  total  within  plot  variance  of  the  non-segregating  families — 
Pi,  P2,  and  Fi.  As  shown  in  Table  10.6,  the  values  obtained  were  25.76  for 
plant  height,  0.71  for  leaf  length,  and  7.38  for  number  of  nodes. 

The  following  symbols  are  used  for  the  components  of  heritable  variance 
(total  phenotypic  minus  environmental) :  alb  =  variance  depending  on  addi- 
tive gene  effects,  a|)  =  variance  depending  on  dominance.  The  heritable 
variance  of  the  F2  was  calculated  and  equated  to:  1/2o-g  +  l/4o-|).  The 
pooled  heritable  variance  of  the  two  first  backcrosses  was  equated  to 
l/2crG  +  l/2o-z).  Solving  for  0%,  the  values  obtained  were  67.32  for  plant 
height,  1.04  for  leaf  length,  and  8.16  for  number  of  nodes.  Values  for  0%, 
as  calculated  by  substitution,  were  113.20  for  plant  height,  0.22  for  leaf 
length,  and  2.20  for  number  of  nodes. 

In  view  of  the  influence  on  node  number  of  a  differential  interaction  of  the 
two  parental  genotypes  with  environment,  an  additional  way  of  approach- 
ing an  analysis  of  the  data  on  this  character  was  tried.  If  a  simple  relation 


FIXING  TRANSGRESSIVE  VIGOR  IN   NICOTIANA  RUSTICA  169 

between  the  environmental  variances  of  the  Pi,  P2,  and  Fi  is  assumed,  so 
that  a%  of  the  Fi  =  1/2((t%  of  Pi  +  al  of  P2),  then  a%  of  the  Fi  =  6.78. 
The  environmental  variance  of  Bi  may  then  be  equated  to  1/2  (variance  of 
Pi+ variance  of  Fi),  which  is  5.12.  By  a  similar  relation,  the  environmental 
variance  of  B2  is  equal  to  8.44.  The  pooled  heritable  variance  of  Bi  +  B2,  i.e., 
l/2ah  +  1/2(7?,,  may  be  equated  to:  (W.49  -  5.12)  +  (9.45  -  8.44).  This 
gave  6.38.  The  heritable  variance  of  the  F2,  i.e.,  l/2ao  +  l/4o-i),  may  be 
equated  to  (10.52  —  6.78).  This  gave  3.74.  Solving:  <t%  =  10.56  and  <to  = 
2.20.  The  former,  (xjj,  has  a  somewhat  larger  value  than  that  obtained  by 
the  original  analysis  (8.16,  Table  10. 6j;  the  latter,  cc,  is  the  same. 

Heritability  of  a  character  was  estimated  as  the  ratio,  expressed  in  per 
cent,  of  the  variance  component  due  to  additive,  fixable  gene  effects  (aa)  to 
the  sum,  <r% -\-  a]) -\-  a%.  Heritability  of  plant  height  was  calculated  to 
be  54.9  per  cent,  of  leaf  length  11.5  per  cent,  and  of  node  number  12.4  per 
cent. 

Estimates  of  the  number  of  effective  factors  (Ki)  were  made  on  the  as- 
sumptions inherent  in  the  equation  A'l  =  (Pi  —  P^^l^a'a.  The  values  ob- 
tained (Table  10. 6 j  were  0.81  for  plant  height,  1.38  for  leaf  length,  and  0.83 
for  number  of  nodes.  These  estimates  were  undoubtedly  too  low,  due  in  part 
to  non-isodirectional  distributions  of  +  and  —  genes  in  the  parents.  Ex- 
perimental evidence  of  non-isodirectional  distribution  was  afforded  by  the 
fixing  of  transgressive  characteristics  in  inbred  selections  following  hybridiza- 
tion between  varieties.  Some  -\-  genes  were  contributed  by  each  parent,  and 
consequently  could  not  have  been  concentrated  in  one.  Linkage  in  coupling 
phase  and/or  differences  in  magnitude  of  effect  of  the  individual  genes  or 
gene  blocks  might  also  have  contributed  to  the  low  estimates  of  the  number 
of  effective  factors. 

In  the  absence  of  data  on  Fa's,  biparental  progenies,  and  double  back- 
crosses  (Mather,  1949),  the  errors  of  the  estimates  of  0%,  a\,,  and  0%  for  each 
character  were  computed  as  follows.  From  the  eight  replications,  four  means 
were  calculated  by  grouping  replications  1  and  2,  3  and  4,  5  and  6,  and  7 
and  8.  The  standard  error  of  the  four  independent  means  was  then  obtained 
(Table  10.6).  These  errors  are  maximum  estimates  since  there  was  a  pro- 
nounced gradient  of  environmental  effects  from  replication  1  to  replication  8. 

Mather  (1949)  is  in  the  process  of  making  an  extensive  biometrical  genetic 
analysis  of  plant  height  in  a  Xicotiana  rustica  cross,  and  it  was  of  interest  to 
compare  his  published  results  with  corresponding  statistics  presented  in  this 
study.  From  his  data  so  far  reported,  the  average  values  (mean  of  1946  and 
1947)  for  components  of  variance  for  plant  height  are:  9.30  for  0%,  9.25  for 
(To,  and  18.05  for  (7%.  The  heritability  calculated  from  these  estimates  is  44.1 
per  cent.  The  results  reported  in  this  discussion  are  similar  in  that  heritabil- 
ity is  high  and  <!%  has  about  twice  the  value  of  a]}. 


170  HAROLD  H.  SMITH 

Results  of  Selection 

The  result  of  selection  for  tall  plants  with  many,  large  leaves  can  be  seen 
by  comparing  the  means  of  Al  with  those  of  the  parental  and  hybrid  genera- 
tions in  Tables  10.1  and  10.2. 

From  the  1947  data  it  is  evident  that  in  the  F4  generation  of  selection  Al  a 
significant  increase  had  been  obtained  over  the  parents  and  Fi's  in  plant 
height  and  green  weight.  This  was  accompanied  by  a  lengthening  in  time 
required  to  reach  maturity.  With  regard  to  this  latter  character,  it  was  noted 
that  the  average  time  for  reaching  maturity  in  five  of  the  six  Fi's  was  later 
than  the  average  of  their  respective  parents.  This  is  contrary  to  the  usual 
result  in  first  generation  hybrids  of  certain  other  plants,  as  maize  and  toma- 
toes; and,  where  early  maturity  is  an  important  economic  character,  would 
generally  not  be  considered  a  manifestation  of  hybrid  vigor,  at  least  in  a 
"beneficial"  sense. 

The  number  of  leaves  in  selection  Al  was  significantly  higher  (P  <  .05) 
than  in  any  of  the  Fi's,  and  all  but  the  most  vigorous  parent,  Olson  68.  Leaves 
of  the  selection  were  shorter  than  the  parent  with  the  longest  leaves  (Olson 
68),  not  significantly  different  from  the  three  Fi's  that  involved  this  parent, 
and  longer  than  in  the  other  three  parents  and  three  Fi's. 

The  1949  data  (Table  10.2)  corroborated  the  1947  results.  There  was  a 
significant  increase  (P  ^  .01)  in  plant  height  and  in  number  of  nodes  over 
the  two  main  parents  and  their  Fi.  Number  of  nodes,  rather  than  of  leaves, 
was  used  since  it  is  a  more  reliable  criterion  of  the  same  character.  As  in 
1947,  there  was  a  less  marked  effect  of  selection  on  leaf  length,  though  there 
appeared  to  be  an  increase  in  Al  from  the  F4  to  the  Fe.  For  this  character 
the  selection  was  superior  to  34753  and  the  Fi,  but  not  significantly  different 
from  Olson  68,  although  a  close  approach  to  significance  at  the  5  per  cent 
level  of  probability  was  reached. 

The  total  within  plot  variances  of  selection  AKFe)  for  plant  height,  num- 
ber of  nodes,  and  leaf  length  were  in  no  case  significantly  higher  than  for  the 
more  variable  parent.  It  was  deduced,  therefore,  that  the  inbred  selection 
had  reached  relative  homozygosity. 

The  general  conclusions  were  that  an  inbred  selection  had  been  produced 
which  had  increased  plant  height,  more  nodes,  heavier  green  weight,  and  a 
longer  growth  period  than  any  parent  or  Fi.  Length  of  leaf  had  been  main- 
tained at  least  at  the  level  of  the  best  parent  variety. 

It  was  also  noted,  though  no  quantitative  data  were  taken,  that  selection 
Al  had  markedly  less  vigorous  sticker  growth  at  topping  time  than  any  of  the 
other  varieties  or  hybrids.  This  is  an  important  agronomic  character. 

Heritability  and  Gain 
One  of  the  objectives  in  conducting  these  experiments  was  to  attempt  to 
determine  to  what  extent  the  progress  realized  in  actual  selection  experi- 


FIXING  TRANSGRESSIVE  VIGOR  IN  NICOTIANA  RUSTICA  171 

ments  could  be  related  to  the  heritability  of  a  character  as  determined  from 
F2  and  first  backcross  data. 

Results  on  the  three  main  characters  studied  were  similar  in  that  there 
was  no  indication  of  complex  genie  interactions,  and  that  estimates  of  the 
number  of  effective  factors  were  low  and  of  the  same  order  of  magnitude  in 
each.  If  we  wish  to  assume  that  the  "reach"  or  selection  differential  (in  terms 
of  standard  deviations)  was  the  same  for  each  character,  and  this  is  approxi- 
mately correct  though  exact  records  on  this  point  are  lacking,  then  the  gain  (in 
terms  of  standard  deviations)  due  to  selection  should  be  roughly  pro])ortional 
to  the  heritability.  The  gain  was  calculated  as  the  difference  between  the 
mean  of  selection  Al  and  the  mid-parent  value,  divided  by  the  standard 
deviation  of  the  F2  (1949  data.  Table  10.2). 

The  relationships  between  heritabilities  and  gains  can  be  observed  by  com- 
paring the  last  two  columns  in  Table  10.6.  With  regard  to  plant  height  and 
leaf  length,  both  heritability  and  gain  are  higher  in  the  former  character; 
though  the  gain  is  less  in  plant  height  than  would  have  been  anticipated  from 
the  relative  heritabilities.  Some  possible  explanations  for  this  latter  result 
could  be  that  the  selection  differential  for  plant  height  was  lower  than  for  leaf 
length,  that  there  was  a  relatively  more  rapid  reduction  in  heritability,  or 
that  an  approach  to  a  physiological  limit  for  tallness  was  made. 

The  gain  in  node  number  is  disproportionately  high  in  relation  to  its 
heritability.  Some  possible  explanations  for  this  result  could  be  that  the 
selection  differential  was  higher,  that  there  was  a  genetic  correlation  with 
plant  height,  or  that  the  selected  character  was  determined  by  a  preponder- 
ance of  recessive  genes  (see  Fi  result),  and  individual  plants  selected  for  high 
node  number  were  largely  homozygous  for  recessive+genes. 

DISCUSSION 

The  experimental  results  have  shown  that  first  generation  crosses  among 
different  varieties  of  Nicotiana  riistica  exhibit  different  degrees  of  character 
expression  ranging  from  the  smaller  parent  value  to  above  the  larger  parent. 
By  selection  and  inbreeding  it  was  possible  to  develop  an  essentially  true- 
breeding  improved  line  which  exceeded  the  best  Pi  or  Fi  in  most  character- 
istics measured. 

This  same  type  of  result  has  also  been  obtained  in  our  experience  with  the 
commercial  species,  N.  tahacum,  and  it  may  be  generally  characteristic  of 
self-fertilized  plants,  as,  e.g.,  Phaseolus  ■vulgaris  (Malinowski,  1928),  soy- 
beans (Veatch,  1930),  and  Galeopsis  (Muntzing,  1930). 

Crossbreeding 
There  have  been  relatively  few  fundamental  changes  in  the  standard 
domestic  varieties  of  N.  tabacum  over  a  long  period  of  years,  except  for  recent 
development   of  types  resistant  to  destructive  diseases   (Garner,   1946). 


172  HAROLD  H.  SMITH 

Houser  (1911)  originally  suggested  the  use  of  first  generation  intervarietal 
tobacco  hybrids  on  a  commercial  scale  to  increase  yields.  He  presented  breed- 
ing results  on  cigar  filler  types,  dating  back  to  1903,  in  which  the  hybrids 
outyielded  the  parent  types  by  as  much  as  57  per  cent.  Plant  breeders  in 
various  tobacco-growing  areas  of  the  world  have  observed  hybrid  vigor 
among  first  generation  hybrids  of  commercial  varieties  (Ashton,  1946),  and 
have  suggested  its  use  in  practice  to  increase  production.  Currently,  con- 
sideration is  being  given  to  improving  the  yield  of  flue-cured  varieties  by 
this  method  (Patel  et  al.,  1949). 

The  results  of  Hayes  (1912),  Hayes,  East,  and  Beinhart  (1913),  and  East 
and  Hayes  (1912)  showed  that  by  intervarietal  hybridization,  selection,  and 
inbreeding  the  number  of  leaves,  an  important  factor  in  yield  of  tobacco, 
could  be  fixed  at  a  level  exceeding  the  parents  or  Fi.  Regarding  the  use  of 
Fi  hybrids  on  a  commercial  scale,  they  stated  (Hayes,  East,  and  Beinhart, 
1913), 

While  it  is  doubtless  true  that  by  this  method  the  yield  could  be  somewhat  increased, 
the  yield  factor,  for  cigar  wrapper  types  at  least,  is  only  of  secondary  importance  com- 
pared with  quality.  Because  of  the  great  importance  of  quality  it  seems  much  more  reason- 
able to  suppose  that  further  advance  can  be  made  by  the  production  of  fixed  types  which  in 
themselves  contain  desirable  growth  factors,  such  as  size,  shape,  position,  uniformity,  vena- 
tion, and  number  of  leaves,  together  with  that  complex  of  conditions  which  goes  to  make 
up  quaHty,  than  by  any  other  method. 

The  problem  of  producing  higher  yielding  varieties  of  N.  tabacum  with 
acceptable  quality  characteristics  of  the  cured  leaf  remains  today.  Kosmo- 
demjjanskii  (1941)  bred  four  families  from  the  cross  Dubec  44XTrebizond 
1272,  two  Russian  varieties  of  .Y.  tabacum,  which,  he  reported,  were  uniform 
for  morphological  characters  and  flavor  and  maintained  transgression  in 
plant  height  and  number  of  leaves  to  the  F;  generation. 

While  first  generation  hybrids  between  selected  parents  may  be  of  use  as 
a  temporary  measure  to  improve  self- fertilizing  crop  plants,  it  would  appear, 
in  so  far  as  can  be  generalized  from  the  results  on  Nicotiana,  that  production 
of  fixed  types  with  favorable  transgressive  characteristics  offers  a  better  long- 
time solution.  Within  any  one  type  of  tobacco,  such  as  flue-cured,  there  are 
currently  available  a  number  of  high  quality  inbred  varieties  which,  though 
of  similar  phenotype,  may  be  expected  to  differ  by  genes  of  a  multifactorial 
system  affecting  size  characteristics  (Emerson  and  Smith,  1950).  Selections 
from  intervarietal  crosses  may  be  expected,  therefore,  to  yield  fixed  types  of 
increased  size  without  presenting  undue  difficulties  to  the  breeder  attempt- 
ing to  maintain  quality. 

In  order  to  discuss  the  hereditary  basis  for  experimental  results  on  hetero- 
sis and  inbreeding,  current  concepts  of  the  genetic  and  evolutionary  mecha- 
nisms involved  are  briefly  presented.  In  the  evolution  of  naturally  crossbred 
organisms,  mutation  and  selection  result  in  the  accumulation  of  dominant 
favorable  genes,  hidden  deleterious  recessives,  and  alleles  or  complexes  of 


FIXING  TRANSGRESSIVE  VIGOR  IN  NICOTIANA  RUSTICA  173 

linked  polygenes  which  give  heterotic  effects  as  helerozygoles.  Heterosis  is 
explained  genetically  as  due  to  the  accumulated  effect  of  the  favorable  domi- 
nants and/or  coadapted  heterozygous  combinations.  It  is  an  adaptive  evolu- 
tionary phenomenon  (Dobzhansky,  1950). 

Selfing 

In  naturally  selfed  populations  there  are  accumulated,  for  the  most  part, 
favorable  genes  that  are  either  dominant,  recessive,  or  lacking  in  strong 
allelic  interactions.  Dominance  is  of  little  evolutionary  significance,  and 
hence  a  preponderance  of  favorable  dominant  genes  is  not  to  be  expected. 
Furthermore,  there  would  ordinarily  be  no  adaptive  significance  to  favorable 
heterozygous  combinations.  One  possible  exception  is  suggested  by  Brieger's 
(1950)  demonstration  that  "if  survival  values  for  both  homozygotes  should 
be  below  0.5  (compared  to  the  heterozygote  value  of  1.0)  in  selfed  populations, 
a  final  equilibrium  is  reached  with  all  three  genotypes  remaining  in  the 
population."  Such  a  condition  might  have  adaptive  value  in  maintaining 
variability  in  selfed  organisms.  Hybrid  vigor  in  self-pollinated  plants,  in 
view  of  the  above  considerations,  is  a  chance  manifestation,  an  "evolutionary 
accident"  causing  luxuriant  growth  (Dobzhansky,  1950),  and  not  an  adap- 
tive product  of  mutation  and  selection. 

However,  from  published  data  on  crosses  within  selfed  species  of  culti- 
vated plants,  it  appears  that  hybrid  vigor  is  of  frequent  rather  than  chance 
occurrence.  Reported  results  with  flax  (Carnahan,  1947),  wheat  (Harrington, 
1944),  barley  (Immer,  1941),  tomatoes  (Larson  and  Currance,  1944),  egg- 
plants (Odland  and  Noll,  1948),  and  soybeans  (Weiss,  Weber,  and  Kalton, 
1947)  all  demonstrated  that  hybrid  vigor  is  characteristic  of  Fi's.  If  these 
data  constitute  a  representative  sample,  then,  although  hybrid  vigor  is  an 
evolutionary  accident  in  naturally  selfed  species,  it  is  not  a  genetical  accident. 

The  result  may  be  interpreted  genetically  as  follows:  Selfed  species  are 
purged  of  deleterious  genes  by  selection.  Different  varieties  within  the 
species  have  accumulated  different  alleles  all  of  which  control  "non-defec- 
tive," slightly  different  physiological  reactions.  The  combination  of  divergent 
alleles  in  heterozygous  condition  may,  more  frequently  than  not,  act  as  East 
has  suggested  in  a  complementary  manner  to  produce  a  more  efficient  physio- 
logical condition.  This  is  expressed  phenotypically  by  the  hybrid  manifesting 
more  vigorous  growth  than  midway  between  the  homozygotes.  Subsequent 
selection  and  inbreeding,  however,  would  permit  an  accumulation  of  the  most 
favorable  alleles  or  gene  complexes  in  the  homozygous  condition. 

As  a  simplified  schematic  example,  let  us  assume  that  two  varieties.  Pi 
and  P2,  differ  by  three  alleles  or  linked  polygene  complexes:  X'  is  dominant 
and  favorable  for  vigorous  growth,  F^  is  a  favorable  recessive,  and  at  the  Z 
locus  the  product  of  the  heterozygous  condition  is  above  the  mean  of  the 


174  HAROLD  H.  SMITH 

homozygotes.  The  composition  of  parents,  Fi,  and  selected  inbred  is  shown 
below  with  arbitrary  "size"  values  assigned  to  each. 

Pi  =  XKX'  (4)  +  y  V  (4)  +Z'Z'  (2)  =10 

Po  =  X'X^  (2)  +  Y^V  (2)  -\-Z'-Z-'  (6)  =10 

Fi  =  X'X'  (4)  +  I'l  ¥'  (2)  ^Z'Z-  (5)  =11 

sel.  =  X'X'  (4)  +  Y'V  (4)  -^Z-Z'-  (6)  =14 

Although  the  difficulty  in  selecting  superior  inbreds  would  become 
greater  with  increasing  numbers  of  effective  segregating  units,  the  following 
advantages  of  selfed  over  crossbred  systems  would  enhance  the  opportunity 
for  success:  (1)  lack  of  deleterious  recessives,  (2)  less  preponderance  of 
dominant  favorable  alleles,  (3)  homozygous  pairs  of  alleles  are  superior,  as  a 
result  of  an  adaptive  evolutionary  process,  to  heterozygous  combinations. 
Naturally  inbred  organisms  are  products  of  historical  evolutionary  processes 
in  which  harmonious  systems  of  homozygous  loci  have  been  selected  to 
attain  optimum  adaptation.  These  considerations  favor  the  expectancy 
and  practicability  of  obtaining  maximum  advance  through  selection  and 
inbreeding  with  self-fertilized  organisms. 

SUMMARY 

There  were  two  general  purposes  in  conducting  these  experiments:  First, 
to  demonstrate  that  by  selection  following  intervarietal  hybridization  in  a 
self-fertilized  organism,  inbreds  could  be  produced  which  transgressed  the 
character  expression  in  parents  and  Fi;  secondly,  to  investigate  the  relation 
between  estimated  heritability  and  the  actual  results  of  selection. 

An  inbred  selection  of  Nicotiana  rustica  which  transgressed  the  Pi  and  Fi 
characteristics  in  plant  height,  node  number,  and  leaf  length  was  obtained. 
The  heritabilities  for  these  three  characters  were  calculated  to  be  54.9  per 
cent,  12.4  per  cent,  and  11.2  per  cent,  respectively.  The  gains  (in  terms  of 
standard  deviations)  due  to  selection  were  1.74,  2.42,  and  0.91,  respectively. 
Some  possible  explanations  for  the  lack  of  direct  proportionality  between 
heritability  and  gain  were  discussed. 

The  number  of  effective  segregating  factors  for  each  of  the  three  characters 
studied  was  estimated  to  be  of  the  same  order  of  magnitude  and  relatively 
few.  Non-isodirectional  distribution  of  -|-  and  —  genes  in  the  parent  varieties 
contributed  to  an  underestimation  of  this  number. 

Non-allelic  interactions  were  apparently  not  an  important  source  of 
variation,  as  indicated  by  scaling  tests  and  evidence  from  double  cross  means. 

Reasons  for  expecting  greater  advances  by  selection  and  inbreeding,  as 
contrasted  to  the  use  of  first  generation  hybrids,  in  naturally  self-fertilizing 
genetic  systems  were  reviewed. 


PAUL  C.  MANGELSDORF 

Harvard  University 


Chapter   1 1 

Hybridization  in 
the  Evolution  of  Moize 


All  varieties  and  races  of  maize  so  far  studied  prove  upon  inbreeding  to  con- 
tain numerous  heterozygous  loci,  and  all  respond  to  inbreeding  with  a  marked 
decline  in  vigor  and  productiveness.  Since  contemporary  maize  is  both 
heterozygous  and  heterotic,  it  is  probable  that  the  factors  which  have  been 
responsible  for  bringing  about  the  present  conditions  are  also  factors  which 
have  played  an  important,  if  not  the  principal  role,  in  the  evolution  of  maize. 

All  of  the  steps  involved  in  the  evolution  of  maize  are  not  yet  known. 
Archaeological  remains  have  told  us  something  of  the  early  stages  of  maize 
under  domestication,  and  we  can  draw  additional  inferences  about  its  original 
nature  from  its  present-day  characteristics.  Our  knowledge  of  the  nature  and 
extent  of  its  present  variation,  although  far  from  complete,  is  already  sub- 
stantial and  is  growing  rapidly.  By  extrapolating  forward  from  ancient 
maize,  and  backward  from  present-day  maize,  w^e  can  make  reasonably  valid 
guesses  about  some  of  the  intermediate  stages  and  about  some  of  the  evolu- 
tionary steps  which  have  occurred  in  its  history. 

The  earliest  known  archaeological  remains  of  maize,  as  well  as  the  best 
evidence  of  an  evolutionary  sequence  in  this  species,  occur  in  the  archaeo- 
logical vegetal  remains  found  in  Bat  Cave  in  New  Mexico  in  1948.  This  ma- 
terial which  covers  a  period  of  approximately  three  thousand  years  (from 
about  2000  B.C.  to  a.d.  1000)  has  been  described  by  Mangelsdorf  and  Smith 
(1949).  It  reveals  three  important  things:  (1)  that  primitive  maize  was  both 
a  small-eared  pop  corn  and  a  form  of  pod  corn;  (2)  that  there  was  an  intro- 
gression  of  teosinte  into  maize  about  midway  in  the  sequence;  (3)  that  there 
was  an  enormous  increase  in  the  range  of  variation  during  the  period  of  ap- 
proximately three  thousand  years  resulting  from  teosinte  introgression  and 
interracial  hybridization. 

175 


176  PAUL  C.  MANGELSDORF 

INTERRACIAL  HYBRIDIZATION  IN  MAIZE 

For  additional  evidence  on  interracial  hybridization  in  maize  we  may 
turn  to  existing  races  of  maize.  Among  these  the  Mexican  races  are  of  par- 
ticular interest  and  significance,  not  because  maize  necessarily  originated  in 
Mexico,  since  there  is  considerable  evidence  that  it  did  not,  but  because 
Mexico  is  a  country  where  primitive  races,  which  in  other  places  are  to  be 
found  primarily  as  archaeological  remains,  still  exist  as  living  entities.  It 
is  possible  in  Mexico  to  find  all  stages  between  ancient  primitive  races  and 
modern  highly-developed  agricultural  races.  One  has  only  to  place  these 
racial  entities  in  their  proper  sequence  in  order  to  have  at  least  the  outline 
of  an  evolutionary  history. 

Wellhausen  et  al.  (1951)  have  recently  made  a  comprehensive  study  of  the 
races  of  maize  of  Mexico.  They  recognize  twenty-five  distinct  races  as  well 
as  several  additional  entities  which  are  still  somewhat  poorly  defined,  but 
some  of  which  may  later  be  described  as  races.  They  divide  the  known  races 
into  four  major  groups  as  follows: 

Group  No.  Races 

1.  Ancient  Indigenous 4 

2.  Pre-Columbian  Exotic 4 

3.  Prehistoric  Mestizos 13 

4.  Modern  Incipient 4 

Origin  of  Mexican  Races  of  Maize 

Ancient  Indigenous  races  are  those  which  are  believed  to  have  arisen  in 
Mexico  from  the  primitive  pod-pop  corn  similar  to  that  whose  remains  were 
found  in  Bat  Cave  in  New  Mexico.  The  races  in  this  group  are  called  in- 
digenous not  because  they  necessarily  had  their  primary  origin  in  Mexico, 
but  because  they  are  thought  to  be  the  product  of  indigenous  differentiation 
from  a  remote  common  ancestor.  The  differentiation  is  assumed  to  have  re- 
sulted from  independent  development  in  different  localities  and  environ- 
ments with  hybridization  playing  little  if  any  part. 

Four  races  of  the  Ancient  Indigenous  group — Palomero  Toluqueno,  Arro- 
cillo  Amarillo,  Chapalote,  and  Nal-tel — are  recognized.  All  of  these,  like  their 
primitive  ancestor,  are  pop  corn.  Two  of  the  four — Chapalote  and  Nal-tel — 
are  forms  of  pod  corn.  All  have  small  ears,  and  all  are  relatively  early  in 
maturity. 

Pre-Columbian  Exotic  races  are  those  which  are  believed  to  have  been 
introduced  into  Mexico  from  Central  or  South  America  before  1492.  Four  of 
these  races — Cacahuazintle,  Harinoso  de  Ocho,  Oloton,  and  Maiz  Dulce — 
are  recognized.  The  evidence  for  their  antiquity  and  exoticism  derives  prin- 
cipally from  two  sources:  all  have  South  American  counterparts;  all  except 
Maiz  Dulce  have  been  parents  of  hybrid  races,  some  of  which  are  them- 
selves relatively  ancient. 


HYBRIDIZATION  IN  THE  EVOLUTION  OF  MAIZE  177 

Prehistoric  Mestizos,^  thirteen  in  number,  are  races  which  are  believed  to 
have  arisen  through  hybridization  between  Ancient  Indigenous  races  and 
Pre-Columbian  races  and  hybridization  of  both  with  a  new  entity,  teosinte. 
The  term  })rehistoric  rather  than  pre-Columbian  is  used  for  this  grouj)  be- 
cause, although  all  are  prehistoric  in  the  sense  that  there  is  no  historical  evi- 
dence of  their  origin,  it  is  not  certain  that  all  are  pre-Columbian. 

Modern  Incipient  races  are  those  which  have  come  into  existence  in  the 
post-Columbian  period.  These  races,  of  which  four  are  recognized,  have  not 
yet  reached  a  state  of  genetic  equilibrium.  They  are  recognizable  entities  but 
are  still  changing. 

The  seventeen  races  comprising  the  two  last  groups  all  appear  to  be  prod- 
ucts of  hybridization,  either  between  races  in  the  first  two  groups,  or  between 
these  races  and  teosinte.  In  several  cases,  secondary  and  even  tertiary  hy- 
bridization seems  to  have  occurred. 

That  a  race  is  the  product  of  previous  hybridization  seems  highly  prob- 
able when  the  following  four  kinds  of  evidence  are  available. 

1.  The  race  is  intermediate  between  the  two  putative  parents  in  a  large 
number  of  characteristics. 

2.  The  putative  parents  still  exist  and  have  geographical  distributions 
which  make  such  hybridization  possible  and  plausible. 

3.  Inbreeding  of  the  suspected  hybrid  race  yields  segregates  which  ap- 
proach in  their  characteristics  one  or  the  other  of  the  two  putative  parents — 
in  some  cases  both. 

4.  A  population  quite  similar  to  the  race  in  question  can  be  synthesized 
by  hybridizing  the  two  putative  parents. 

Wellhausen  el  al.  (1951)  have  presented  all  four  kinds  of  evidence  for  the 
hybrid  origin  of  a  number  of  the  present-day  Mexican  races.  They  have  pre- 
sented similar  but  less  complete  evidence  for  the  remainder. 

The  variety  Conico,  for  example,  which  is  the  most  common  race  in  the 
Valley  of  Mexico,  is  clearly  the  product  of  hybridizing  the  ancient  Palomero 
Toluqueno  with  the  exotic  Cacahuazintle.  Conico  is  intermediate  between 
these  two  races  in  many  characteristics.  The  two  putative  ancestral  races  still 
are  found  in  isolated  localities  in  the  Valley  of  Mexico.  The  race  is  interme- 
diate in  its  characteristics  between  the  two  suspected  parents.  Inbreeding 
yields  segregates  which  almost  duplicate  in  their  characteristics  one  of  the 
parents — Palomero  Toluqueno.  Segregates  approaching  the  other  suspected 
parent,  Cacahuazintle,  also  result  from  inbreeding  but  this  parent  is  never 
exactly  duplicated.  Obviously  the  race  has  become  something  more  com- 
plex than  a  mixture  of  equal  parts  of  two  earlier  races.  Nevertheless  the 
crossing  of  Palomero  Toluqueno  and  Cacahuazintle  still  produces  a  hybrid 
which  in  many  respects  is  scarcely  distinguishable  from  the  suspected  hybrid 
race.  The  data  in  Table  11.1  show  that  Conico  is  intermediate  between  Palo- 

1.  Mestizo  is  the  Latin-American  term  for  a  racial  hybrid. 


178 


PAUL  C.  MANGELSDORF 


mero  Toluqueno  and  Cacahuazintle  in  a  large  number  of  characteristics.  They 
also  show  how  closely  a  recently-made  hybrid  of  these  two  ancient  races  re- 
sembles the  suspected  hybrid  race,  Conico.  Ears  of  the  three  races  and  the 
hybrid  are  illustrated  in  Figure  11.1. 

The  hybrid  race,  Conico,  has  in  turn  been  the  ancestor  of  still  more  complex 
hybrid  races.  A  Modern  Incipient  race,  Chalqueho,  which  has  originated  in  his- 
torical times  in  the  vicinity  of  the  village  of  Chalco  in  the  Valley  of  Mexico, 

TABLE  11.1 
COMPARISON  OF  CONICO  WITH  ITS  PUTATIVE  PARENTS* 


Characters 


Ears  and  plants: 

Ear  diameter,  mm 

No.  rows  grain 

Width  kernels,  mm 

Thickness  kernels,  mm 

Diameter  peduncle,  mm 

Length  ear,  cm 

Height  plant,  cm 

Tillering  index 

Pilosity  score 

Internal  ear  characters: 

Ear  diameter,  mm 

Cob  diameter,  mm 

Rachis  diameter,  mm 

Length  kernels,  mm 

Estimated  rachilla  length,  mm 

Cob/rachis  index 

Glume/kernel  index 

Rachilla/kernel  index 

Pedicel  hairs  score 

Rachis  flap  score 


Races 


Palomero 
Toluqueno 


37.1 
21.8 
4.6 
2.8 
8.0 
9.8 
175 
.26 
3 


34.0 
19.5 
10.4 
11.4 
.4 
1.88 


.40 
.04 


0 
0 


Fi 
Hybrid 


Conico 


45.2 
18.6 
6.8 
3.6 
9.2 
11.8 
200 
.35 
4 


45.1 

15.7 

7.4 

3.9 

9.8 

12.6 

193 

.22 
3-4 


42.4 

19.0 

9.6 

14.8 
1.6 
1.98 
.32 
.11 
2-4 


2-3 


Cacahua- 
zintle 


53.2 
16.2 
9.8 
5.3 
10.6 
14.7 
210 
.39 
4 


47.0 
27.7 
11.7 
14.0 

3.6 

2.37 
.57 
.26 

4 

3 


*  After  Wellhausen  el  al. 


is  undoubtedly  the  product  of  hybridizing  Conico  with  Tuxpeiio,  a  pro- 
ductive lowland  race  of  the  Prehistoric  Mestizo  group.  Since  Tuxpeno  is 
itself  a  hybrid,  the  postulated  pedigree  for  Chalqueno,  which  is  shown  in 
Figure  11.2,  becomes  quite  complex. 

In  the  pedigree  of  Tuxpeno  a  distinction  has  been  made  (by  employing 
different  styles  of  type)  between  the  facts  which  are  well-established  and 
those  which  are  largely  based  upon  inference.  There  is  little  doubt  that 
Conico  is  a  hybrid  of  Palomero  Toluqueno  and  Cacahuazintle,  or  that  Chal- 
queno is  a  hybrid  of  Conico  and  Tuxpeno.  There  is  little  doubt  that  Tuxpeno 
is  a  hybrid  derivative  of  Tepecintle,  but  it  is  not  certain  that  the  other  par- 
ent is  Olotillo,  although  this  is  the  best  guess  which  can  be  made  with  the 


Fig.  11.1 — Ears  of  the  Mexican  maize  races  Palomero  Toluqueno,  Conico,  and  Cacahua- 
cintle.  Conico  is  intermediate  between  the  two  other  races  and  is  thought  to  be  the  product 

of  their  hybridization. 


CHALQUENO 


CONICO 


PALOMERO    TOLUQUENO 


"•CACAHUACINTLE 


TUXPENO 


OLOTILLO 


^TEPECINTLE 


HARINOSO    FLEXIBLE 


TEOSINTE 


HARINOSO    DE   GUATEMALA 


^TEOSINTE 


Fig.  11.2— The  postulated  geneology  of  the  Mexican  race  Chalqueno.  Parts  of  the  geneal- 
ogy not  well  established  by  experimental  evidence  are  shown  in  Italics. 


180  PAUL  C.  MANGELSDORF 

evidence  now  at  hand.  That  Olotillo  and  Tepecintle  are  both  hybrid  races 
involving  teosinte  is  even  more  difficult  to  prove,  although  data  on  chromo- 
some knobs  presented  by  Wellhausen  et  al.  tend  to  substantiate  such  a  con- 
clusion. 

There  is  at  least  no  doubt  that  interracial  hybridization  has  been  an  im- 
portant factor  in  the  evolution  of  maize  in  Mexico.  Has  this  hybridization 
produced  heterosis,  or  has  it  merely  resulted  in  Mendelian  recombination? 
The  extent  to  which  the  suspected  hybrid  races  remain  intermediate  be- 
tween the  two  putative  parents  suggests  that  natural  selection  (operating  in 
a  man-made  environment)  has  tended  to  preserve  the  heterozygote  and  to 
eliminate  the  segregates  which  approach  homozygosity.  It  is  at  least  certain 
that  the  hybrid  races  are  intermediate  between  their  putative  parents  in 
their  characteristics  to  a  remarkable  degree  and  that  they  are  highly  hetero- 
zygous. Even  in  the  absence  of  natural  selection  favoring  the  more  heterozy- 
gous individuals,  there  would  seem  to  be  a  tendency  for  repeated  interracial 
hybridization  to  create  an  ever-increasing  degree  of  heterosis.  This  is  the 
consequence  of  the  fact  that  maize  is  a  highly  cross-pollinated  plant,  and 
that  heterozygosity  does  not  diminish  after  the  Fo  in  cross-fertilized  popula- 
tions in  which  mating  is  random. 

Wright  (1922)  has  suggested  that  the  vigor  and  productiveness  of  an  Fo 
population  falls  below  that  of  the  Fi  by  an  amount  equal  to  l/«  of  the  dif- 
ference between  the  production  of  the  Fi  and  the  average  production  of  the 
parental  stock,  where  n  is  the  number  of  inbred  strains  which  enter  into  the 
ancestry  of  a  hybrid.  The  formula  is  also  applicable  to  hybrids  in  which  the 
parental  stocks  are  not  inbred  lines,  but  are  stable  open-pollinated  varieties 
in  which  random  mating  does  not  diminish  vigor.  It  is,  of  course,  not  ap- 
plicable to  hybrids  of  single  crosses  which  are  themselves  subject  to  dimin- 
ished vigor  as  the  result  of  random  mating. 

Hybrid  Vigor  in  Advanced  Generations 

The  rate  at  which  hybrid  vigor  diminishes  in  a  population  after  the  F2  gen- 
eration is  related  to  the  proportion  of  outcrossing.  This  is  true  whether  hybrid 
vigor  depends  upon  heterozygosity  or  upon  the  cumulative  action  of  dominant 
genes,  and  irrespective  of  the  number  of  genes  involved  and  the  degree  of 
linkage.  With  complete  selffiig  the  amount  of  hybrid  vigor  retained  is  halved 
in  each  succeeding  generation.  With  complete  outcrossing  the  amount  of 
hybrid  vigor  falls  to  one-half  in  the  F2  and  thereafter  remains  constant.  With 
a  mixture  of  selfing  and  outcrossing  an  intermediate  result  is  to  be  expected. 
This  can  be  calculated  from  the  following  formula  presented  by  Stephens 

(1950): 

h  =  l[{\-k)h'-Vk]. 

In  this  formula  h  is  the  proportion  of  Fi  vigor  retained  in  the  current  gen- 
eration, //  is  the  proportion  retained  in  the  preceding  generation,  and  k  is 


HYBRIDIZATION   IN  THE  EVOLUTION   OF  MAIZE  181 

the  proportion  of  outcrossing.  The  formula  is  based  upon  the  assumption  that 
gene  action  is,  on  the  average,  additive. 

It  is  obvious  (according  to  this  formula)  that  the  percentage  of  hybrid 
vigor  retained  in  later  generations  of  a  cross  will  approach  but  never  fall  be- 
low kjl.  Since  the  value  of  k  in  the  case  of  maize  lies  usually  between  .9  and 
1.0,  it  is  apparent  that  the  amount  of  hybrid  vigor  retained  in  later  genera- 
tions of  maize  crosses  will  (with  random  mating)  seldom  fall  below  the 
one-half,  which  is  characteristic  of  the  Fs- 

There  are  experimental  data  which  tend  to  show  that  advanced  genera- 
tions of  maize  crosses  behave  approximately  as  would  be  expected  from  the 
formulae  of  Wright  and  Stephens. 

Kiesselbach  (1930)  compared  the  Fi,F2,andF3of  21  single  crosses  with  the 
parental  inbred  lines.  The  average  yield  of  the  inbreds  was  24.0  bushels.  The 
average  yield  of  the  Fi  was  57.0  bushels.  The  theoretical  yield  of  the  Fo  is 
40.5  bushels.  The  actual  yield  was  38.4  bushels  which  does  not  differ  signifi- 
cantly from  the  theoretical.  The  yield  of  the  F3  was  37.8  bushels  which  is 
almost  identical  to  the  F2  yield. 

Neal  (1935)  compared  the  yield  in  Fi  and  F2  of  10  single  crosses,  4  three- 
way  crosses,  and  2  double  crosses.  The  theoretical  reduction  in  yield  be- 
tween the  Fi  and  F2  in  these  three  groups  (based  upon  Wright's  formula) 
should  have  been  31.1  per  cent,  21.0  per  cent,  and  15.2  per  cent  respectively. 
The  actual  reduction  was  29.5  per  cent,  23.4  per  cent,  and  15.8  per  cent.  The 
agreement  could  scarcely  have  been  closer. 

There  is  abundant  evidence  from  maize  crosses  to  show  that  equilibrium 
is  reached  in  F2,  and  that  in  the  absence  of  selection  there  is  no  further  reduc- 
tion in  yield  in  the  F3.  Data  from  the  experiments  of  Kiesselbach  (1930), 
Neal  (1935),  and  Sprague  and  Jenkins  (1943)  are  summarized  in  Table  11.2. 

The  data  so  far  presented  are  concerned  with  crosses  of  inbred  strains.  Do 
hybrids  of  open-pollinated  varieties  behave  in  the  same  w^ay?  Since  open- 
pollinated  varieties,  although  not  homogeneous,  are  stable  in  productiveness 
they  should  behave  in  crosses  in  the  same  way  as  inbred  strains.  Data  from 
advanced  generations  of  topcrosses  presented  by  Wellhausen  and  Roberts 
(1949)  indicate  that  they  do.  The  theoretical  yields  of  the  F2  of  a  topcross 
can  be  computed  from  a  formula  suggested  by  Mangelsdorf  (1939). 

W'ellhausen  and  Roberts  compared  the  Fi  and  F2  generations  of  31  dif- 
ferent topcrosses  each  including  the  open-pollinated  variety  Urquiza  and 
two  inbred  lines  of  unrelated  varieties.  The  latter  were  in  all  cases  first-gener- 
ation selfs.  The  mean  yield  of  the  31  Fi  hybrids  (in  terms  of  percentage  of 
Urquiza)  was  132  per  cent.  The  mean  yield  of  the  corresponding  31  F2  hy- 
brids was  126  per  cent.  Since  the  yields  of  the  first-generation  selfed  lines 
entering  into  the  cross  is  not  known,  it  is  impossible  to  calculate  with  pre- 
cision the  theoretical  yield  of  the  F2.  However,  it  is  known  that  good  homozy- 
gous inbreds  yield  approximately  half  as  much  as  open-pollinated  varieties 


182 


PAUL  C.  MANGELSDORF 


(Jones  and  Mangelsdorf,  1925;  Neal,  1935)  which  means  that  inbreds,  selfed 
once  and  having  lost  half  of  their  heterozygosity,  should  yield  75  per  cent  as 
much  as  the  open-pollinated  varieties  from  which  they  were  derived.  Assum- 
ing that  the  single-cross  combinations  involved  are  at  least  equal  to  the  top- 
cross  combinations — 132  per  cent — we  compute  the  theoretical  F2  yield  of 
the  topcrosses  at  117  per  cent,  which  is  considerably  less  than  the  126  per 
cent  actually  obtained  in  the  experiments.  From  the  results  it  can  be  con- 
cluded that  hybrid  combinations  including  open-pollinated  varieties  of  maize 
retain  a  considerable  proportion  of  their  vigor  in  advanced  generations. 

There  is  also  some  evidence  to  indicate  that  the  amount  of  heterosis  which 
occurs  when  open-pollinated  varieties  are  used  in  hybrid  combinations  may  be 

TABLE  11.2 

SUMMARY  OF  EXPERIMENTS  DEMONSTRATING  EQUILIBRIUM 
REACHED  IN  Fo  AND  NO  ADDITIONAL  YIELD  REDUC- 
TION IN  F3  OF  MAIZE  CROSSES 


Class  of  Hybrids 

No. 
Hybrids 
Tested 

Yield  in  Per  Cent  of  Fi 

Investigators 

Fi 

F2 

Fs 

Kiesselbach,  1930 

Neal,  1935 

Single  crosses 
Single  crosses 
3-way  crosses 
Synthetics 

21 

10 

4 

5 

100 
100 
100 
100 

68.0 
70.5 
76.6 
94.3 

66.0 

75.7 

Neal,  1935 

Sprague  and  Jenkins,  1943. . . . 

75.8 
95.4 

Total  and  averages 

40 

100 

76.9 

78.2 

considerably  higher  with  Latin-American  varieties  than  with  varieties  com- 
monly grown  in  the  United  States.  Wellhausen  and  Roberts  report  single 
topcrosses  yielding  up  to  173  per  cent  of  the  open-pollinated  variety  and 
double  topcrosses  up  to  150  per  cent.  A  recent  report  from  the  Ministry  of 
Agriculture  of  El  Salvador  (1949)  shows  four  different  hybrids  between  open- 
pollinated  varieties  yielding  about  50  per  cent  more  than  the  average  of  the 
parents.  Such  increases  are  not  surprising,  since  the  varieties  used  in  the 
experiments  are  quite  diverse,  much  more  so  than  Corn  Belt  varieties. 

All  of  the  data  which  are  available  on  the  yields  of  advanced  generations 
of  maize  crosses,  whether  the  parents  be  inbred  strains  or  open-pollinated 
varieties,  tend  to  show  that  a  substantial  part  of  the  hybrid  vigor  charac- 
teristic of  the  Fi  is  retained  in  subsequent  generations.  Thus  maize  under 
domestication  is  potentially  and  no  doubt  actually  a  self-improving  plant. 
Distinct  more-or-less  stable  varieties  or  races  evolve  in  the  isolation  of 
separated  regions.  Man  brings  these  varieties  or  races  together  under  condi- 
tions where  cross-fertilization  is  inevitable,  and  a  new  hybrid  race  is  born. 
Repeated  cycles  of  this  series  of  events  inevitably  lead  to  the  development, 


HYBRIDIZATION  IN  THE  EVOLUTION  OF  MAIZE  183 

without  any  direct  intervention  of  man,  of  more  productive  races.  If,  in  addi- 
tion, natural  selection  favors  the  heterozygous  combinations  as  it  does  in 
Drosophila  (Dobzhansky,  1949),  then  the  retention  of  hybrid  vigor  in  ad- 
vanced generations  of  maize  crosses  will  be  even  greater  than  that  indicated 
by  the  experimental  results. 

INTER-SPECIFIC  HYBRIDIZATION  OF  MAIZE  AND  TEOSINTE 

Superimposed  upon  these  evolutionary  mechanisms,  at  least  in  Me.xico 
and  Central  America,  is  a  second  kind  of  hybridization  which  involves  the 
introgression  of  teosinte  into  maize.  The  importance  of  this  evolutionary 
factor  would  be  difficult  to  overemphasize,  for  as  Wellhausen  el  al.  have 
shown  all  of  the  more  productive  races  of  maize  of  Mexico  show  evidence  of 
past  teosinte  introgression. 

The  genetic  nature  of  teosinte  need  not  enter  into  the  present  discussion.  Dr. 
R.  G.  Reeves  and  I  concluded  some  years  ago  that  teosinte  is  not,  as  many 
botanists  have  supposed,  the  ancestor  of  maize,  but  is  instead  the  progeny 
of  a  cross  of  maize  and  Tripsacum.  This  conclusion  has  not  yet  been  ex- 
perimentally proven,  and  although  there  is  much  evidence  to  support  it,  it  is 
by  no  means  universally  accepted  by  other  students  of  corn's  ancestry.  For 
the  purpose  of  this  discussion  we  need  not  debate  this  particular  point,  since 
we  need  only  to  recognize  that  there  is  a  well-defined  entity  known  as  teo- 
sinte which  occurs  as  a  weed  in  the  corn  fields  of  central  Mexico  and  as  a  wild 
plant  in  Southwestern  Mexico,  Guatemala,  and  Honduras. 

Teosinte  is  far  more  common  than  formerly  supposed.  Twenty -five  years 
ago  its  occurrence  was  known  in  only  three  or  four  localities  in  Mexico.  Since 
then,  numerous  additional  sites  have  been  described  in  Mexico  and  Guate- 
mala, and  recently  a  locality  in  Honduras  has  been  added  (Standley,  1950). 

Teosinte  is  the  closest  relative  of  maize.  It  has  the  same  chromosome  num- 
ber (ten)  as  maize,  and  hybridizes  easily  with  it  to  produce  hybrids  which  are 
completely  fertile,  or  almost  so.  The  chromosomes  of  corn  and  teosinte  are 
homologous  to  the  extent  that  they  pair  almost  completely.  Crossing  over 
between  teosinte  and  corn  chromosomes  is  of  the  same  order  as  crossing  over 
in  pure  corn  (Emerson  and  Beadle,  1932). 

Present-Day  Hybridization 
Since  both  teosinte  and  maize  are  wind-pollinated  plants  and  since  they 
hybridize  easily,  it  is  almost  inevitable  that  hybridization  between  the  two 
species  should  occur  in  any  region  where  both  are  growing.  There  is  no  doubt 
that  such  hybridization  is  constantly  occurring,  and  that  it  has  been  going 
on  for  many  centuries.  Fi  hybrids  of  com  and  teosinte  have  been  collected 
in  both  Mexico  and  Guatemala.  They  are  especially  common  in  Central 
Mexico  where  teosinte  grows  as  a  weed.  In  1943, 1  obtained  some  data  on  the 
extent  to  which  hybridization  occurs  near  the  village  of  Chalco  where  teosinte 
is  a  common  weed  in  and  around  the  corn  fields.  In  a  field  where  teosinte  oc- 


184  PAUL  C.  MANGELSDORF 

curred  abundantly  as  a  weed  permission  was  obtained  from  the  owner  to  tag 
and  harvest  500  consecutive  plants.  Of  the  500  plants  tagged,  288  proved  to 
be  maize,  219  were  teosinte  plants,  and  3  were  Fi  hybrids.  Of  the  288  ears 
classified  as  maize,  4  showed  definite  evidence  of  contamination  with  teosinte 
in  earlier  generations.  In  addition,  one  ear  was  found  in  an  adjacent  row  (not 
part  of  the  sample  of  500  plants)  which  was  identical  in  its  characteristics 
with  a  first  backcross  to  teosinte. 

The  plants  in  this  field  therefore  furnished  unmistakable  evidence  of  hy- 
bridization, both  present  and  during  the  recent  past,  between  maize  and 
teosinte.  One  plant  out  of  every  167  plants  in  the  field  was  a  vigorous  Fi  hy- 
brid shedding  abundant  pollen  which  became  part  of  the  general  pollen  mix- 
ture in  the  field.  The  Fi  hybrids  themselves,  in  spite  of  their  vigor,  have  a  low 
survival  value.  The  Mexican  farmer  makes  no  distinction  between  teosinte 
and  the  Fi  hybrids.  Both  are  left  standing  in  the  field  when  the  corn  is  har- 
vested. The  pure  teosinte  disperses  its  seeds  which  are  enclosed  in  hard  bony 
shells,  and  a  new  crop  of  teosinte  plants  appears  the  following  spring.  But 
the  Fi  hybrids  have  no  effective  means  of  seed  dispersal,  and  their  seeds,  only 
partially  covered,  are  quite  vulnerable  to  the  ravages  of  insects  and  rodents. 

Both  maize  and  teosinte  are  quite  successful  in  occupying  distinct  niches 
in  Mexican  corn  fields.  The  one,  a  cultivated  plant,  depends  for  its  survival 
upon  its  usefulness  to  man.  The  other,  a  weed,  depends  for  survival  upon  its 
well-protected  kernels  and  its  efficient  method  of  dispersal.  There  is  no  such 
niche  for  the  Fi  hybrid.  It  is  discarded  by  man  as  a  cultivated  plant,  and  it 
cannot  compete  with  teosinte  as  a  weed.  "Finding  no  friend  in  either  nature 
or  man"  (to  use  Weatherwax's  apt  description)  the  Fi  hybrids  would  be  of 
no  evolutionary  significance  were  it  not  for  the  fact  that  they  hybridize  with 
both  parents.  Thus  there  is  a  constant  introgression  of  teosinte  into  maize  and 
of  maize  into  teosinte.  In  the  vicinity  of  Chalco,  in  Mexico,  this  process  has 
gone  on  so  long  and  the  teosinte  has  become  so  maize-like  in  all  of  its  charac- 
ters, that  maize  and  teosinte  plants  can  no  longer  be  distinguished  until  after 
the  pistillate  inflorescences  have  developed.  The  teosinte  of  Chalco  has  "ab- 
sorbed" the  genes  for  hairy  leaf  sheaths  and  red  color  characteristic  of  the 
maize  of  the  region.  Individual  plants  of  teosinte  have  been  found  which  have 
the  yellow  endosperm  color  of  corn,  although  teosinte  is  normally  white- 
seeded. 

The  introgression  of  teosinte  into  maize  in  Mexico  today  is  an  established 
fact.  The  question  is  how  long  this  process  has  been  going  on  and  whether  it 
is  strictly  a  local  phenomenon  or  whether  it  has  affected  the  maize  varieties 
of  America. 

Practically  all  students  of  maize  and  its  relatives  recognize  that  teosinte 
varieties  differ  in  the  degree  to  which  they  have  become  maize-like.  Longley 
(1941),  for  example,  considers  the  teosinte  of  Southern  Guatemala  to  be  the 
least  maize-like  and  that  of  Mexico  the  most  maize-like. 


HYBRIDIZATION   IN  THE   EVOLUTION   OF  MAIZE  185 

Rogers  (1950)  has  shown  that  teosinte  varieties  differ  quite  markedly  in 
their  genes  governing  the  characteristics  in  which  maize  and  teosinte  differ, 
especially  characters  of  the  j)istillate  inflorescence,  tillering  habit,  and  re- 
sponse to  length  of  day.  He  attributes  these  differences  to  varieties  in  the 
type  and  amount  of  maize  germplasm  which  has  become  incorjiorated  into 
teosinte. 

If  teosinte  varieties  differ  in  the  amount  and  kind  of  maize  contamination 
which  they  now  contain,  it  is  difficult  to  escape  the  conclusion  that  maize 
varieties  must  likewise  differ  in  the  amount  of  teosinte  contamination.  There 
is  little  doubt  that  maize  varieties  do  differ  in  this  respect. 

Ancient   Hybridization 

The  prehistoric  maize  from  Bat  Cave  has  already  been  briefly  mentioned. 
The  earliest  Bat  Cave  corn,  dated  at  approximately  2000  B.C.,  shows  no 
evidence  whatever  of  teosinte  introgression.  Beginning  about  midway  in  the 
series  (which  would  be  about  500  b.c.  if  the  sequence  were  strictly  linear  but 
which,  according  to  unpublished  radio-carbon  determinations  made  by  Libby , 
is  probably  somewhat  later)  cobs  make  their  appearance  which  are  scarcely 
distinguishable  from  the  cobs  which  we  have  produced  experimentally  by 
crossing  corn  and  teosinte.  \\'eatherwax  (1950)  regards  this  evidence  of  teo- 
sinte introgression  as  far  from  conclusive,  and  it  is,  of  course,  quite  impossible 
to  prove  that  a  cob  a  thousand  years  or  more  old  is  the  product  of  hybridiza- 
tion of  maize  and  teosinte.  Nevertheless,  it  is  true  that  teosinte  introgression 
produces  certain  definite  effects  upon  the  cob,  as  some  of  us  who  have  studied 
the  derivatives  of  teosinte-maize  crosses  on  an  extensive  scale  are  well  aware. 

When  it  is  possible  to  duplicate  almost  exactly  in  experimental  cultures 
specimens  found  in  nature,  the  odds  are  at  least  somewhat  better  than  even 
that  the  resemblance  between  the  two  specimens  is  more  than  coincidence. 
There  is  little  doubt  in  my  mind  that  the  later  Bat  Cave  corn  is  the  product 
of  contamination  with  teosinte.  Certainly  it  differs  from  the  earlier  Bat  Cave 
corn  quite  strikingly,  and  it  is  exactly  the  way  in  which  it  would  be  expected 
to  differ  if  it  is  the  product  of  teosinte  introgression. 

Significance   of  Chromosome   Knobs 

Mangelsdorf  and  Reeves  (1939)  suggested  some  years  ago  that  the  deeply 
staining  heterochromatic  knobs,  characteristic  of  the  chromosomes  of  many 
varieties  of  maize,  are  the  result  of  the  previous  hybridization  of  maize  and 
teosinte,  or  more  remotely  of  maize  and  Tripsacum.  There  has  been  much 
indirect  evidence  in  support  of  this  hypothesis  (especially  Mangelsdorf  and 
Cameron,  1942;  Reeves,  1944),  and  the  recent  studies  of  Wellhausen  ei  al. 
on  Mexican  races  of  maize  provide  additional  evidence  of  this  nature. 
Chromosome  knob  number  in  Mexican  races  is  closely  correlated  with  the 
characteristics  of  the  races.  The  four  Ancient  Indigenous  races,  assumed  to 
be  relatively  pure  corn,  have  an  average  chromosome  knob  number  of  4.2. 


186 


PAUL  C.  MANGELSDORF 


The  four  Pre-Columbian  Exotic  races,  also  believed  to  be  relatively  free 
from  contamination,  have  an  average  chromosome  knob  number  of  4.3.  The 
thirteen  Prehistoric  Mestizos  and  the  four  Modern  Incipient  races  (all  except 
one  of  which  are  assumed  to  involve  teosinte  introgression)  have  chromo- 
some knob  numbers  of  7.1  and  8.0,  respectively. 

It  is  interesting  to  note  that  in  races  for  which  hybridization  is  postulated 
the  hybrid  race,  although  usually  intermediate  in  chromosome  knob  number 
between  its  two  putative  parents,  resembles  most  closely  the  parent  with  a 
high  knob  number.  For  the  eleven  hybrid  races  for  which  chromosome  knob 
numbers  are  available,  not  only  for  the  hybrid  races  but  for  the  two  suspected 
parent  races,  the  data  (Table  11.3)  are  as  follows:  the  average  of  the  lower- 

TABLE  11.3 

CHROMOSOME  KNOB  NUMBERS  OF  MEXICAN  HY- 
BRID RACES  OF  MAIZE  AND  OF  THEIR 
PUTATIVE  PARENTS* 


Hybrid  Race 


Tabloncillo 

Comiteco 

Jala 

Zapalote  Chico. . 
Zapalote  Grande 

Tuxpeno 

Vandeiio 

Chalqueno 

Celaya 

Conico  Norteno. 
Bolita 

Averages . .  . 


Race 


7.6 
5.6 
7.5 
11.7 
7.4 
6.1 
8.1 
6.8 
8.5 
8.0 
8.6 


7.8 


Parents 


Lower 


low 
5.0 
5.6 
5.5 
7.0 
6.3 
6.1 
1.0 
6.1 
1.0 
7.6 


5.1 


Higher 


8.0 


.0 
.6 
.0 

,7 
.0 


7. 
7. 
9. 

11. 
9. 
7.4 
6.1 
7.6 
8.5 

11.7 


8.5 


*  Data  from  Wellhausen  et  al. 

numbered  parent  was  5.1  knobs,  of  the  higher-numbered  parent,  8.5  knobs, 
of  the  hybrid,  7.8  knobs.  The  fact  that  the  average  knob  number  in  the 
hybrid  races  approaches  the  average  knob  number  of  the  higher  parents 
suggests,  perhaps,  that  natural  selection  has  tended  to  retain  the  maximum 
amount  of  teosinte  introgression  and  hence  the  maximum  number  of  knobs. 

The  Effects  of  Hybridizing  Maize  and  Teosinte 
There  is  no  doubt  that  maize  and  teosinte  are  hybridizing  in  Mexico  and 
Central  America  today,  and  there  is  at  least  a  strong  indication  that  they 
have  done  so  in  the  past.  What  have  been  the  effects  of  that  hybridization? 
One  valid  way  of  determining  what  happens  when  teosinte  introgresses 
into  maize  is  to  produce  such  introgression  experimentally.  This  has  been 
done  on  an  extensive  scale  by  crossing  an  inbred  strain  of  maize,  Texas  4R-3, 


HYBRIDIZATION  IN  THE  EVOLUTION  OF  MAIZE  187 

with  four  varieties  of  teosinte,  and  by  repeatedly  backcrossing  (three  times 
in  most  instances)  the  hybrids  to  the  inbred  strain,  retaining  various  amounts 
of  teosinte  germplasm  through  selection.  The  end  result  is  a  series  of  modified 
inbred  strains  approximately  like  the  original  4R-3 — all  relatively  isogenic 
except  that  parts  of  one  or  more  chromosome  segments  from  teosinte  have 
been  substituted  for  homologous  parts  from  maize. 

That  the  substitution  involves  chromosome  segments  or  blocks  of  genes 
and  not  single  genes  is  strongly  indicated  by  the  fact  that  the  units  have 
multiple  effects  and  that  there  is  breakage  within  the  units  in  some  cases, 
although  in  general  they  are  transmitted  intact.  Their  mode  of  inheritance 
and  their  linkage  relations  can  be  determined  as  though  they  were  single 
genes.  Yet  each  of  the  units  affects  many  if  not  all  of  the  characters  in  which 
maize  and  teosinte  differ.  The  block  of  genes  on  chromosome  3,  for  example, 
although  inherited  intact  as  a  single  hereditary  unit,  affects  number  of  ears, 
size  of  ear,  number  of  seeds,  size  of  seeds,  number  of  rows  of  grain,  staminate 
spikelets  on  the  ear,  and  induration  of  the  rachis.  In  addition  it  has  a  con- 
cealed effect,  discussed  later,  upon  such  characters  as  response  to  length  of 
day  and  the  development  of  single  spikelets.  The  block  of  genes  on  chromo- 
some 4  has  practically  the  same  effects  in  somewhat  greater  degree,  but  this 
block  shows  definite  evidence  of  breakage  or  crossing  over  which  is  of  the 
order  of  30  per  cent. 

These  blocks  of  genes  are  not  random  samples  of  teosinte  germplasm,  but 
represent  definite  genie  entities  which  are  transmitted  from  teosinte  to  maize 
in  the  process  of  repeated  backcrossing.  Different  varieties  of  teosinte  yield 
comparable  if  not  identical  blocks  of  genes,  and  the  same  variety  of  teosinte 
in  different  crosses  does  likewise.  Regardless  of  the  amount  of  introgression 
of  maize  which  teosinte  has  undergone  in  its  past  history,  and  regardless  of 
the  differentiation  which  has  occurred  between  varieties  of  teosinte,  there  are 
still  regions  in  all  varieties  of  teosinte,  perhaps  near  the  centromeres,  which 
have  remained  "pure"  for  the  original  genes. 

Effects  in  Heterozygous  Condition 
When  these  blocks  of  genes  are  introduced  into  maize  they  have  profound 
effects  which  differ  greatly  in  the  heterozygous  and  homozygous  condition. 
Since  maize  and  teosinte  represent  completely  different  morphological  and 
physiological  systems  (especially  from  the  standpoint  of  their  pistillate  in- 
florescences and  their  response  to  length  of  day),  this  substitution,  of  seg- 
ments of  chromatin  from  one  species  for  homologous  segments  from  the 
other,  represents  a  drastic  interchange  of  parts  comparable,  perhaps,  to  in- 
stalling a  carburetor  or  other  essential  part  from  one  make  of  car  into  an- 
other. In  the  Fi  hybrid  of  corn  and  teosinte  where  the  blocks  of  genes  are 
heterozygous,  there  is  no  particular  functional  difficulty.  Here  the  two  com- 
plete systems  are  operating  simultaneously  and  the  result  is  a  vigorous  hy- 


188 


PAUL  C.  MANGELSDORF 


brid,  vegetatively  luxuriant,  potentially  capable  of  producing  great  numbers 
of  seed.  Measured  solely  by  total  grain  yield,  the  Fi  hybrid  does  not  exhibit 
heterosis  since  its  grain  yield  is  considerably  less  than  that  of  corn,  but  meas- 
ured in  terms  of  number  of  seeds,  or  number  of  stalks,  or  total  fodder,  the 
hybrid  certainly  exhibits  heterosis. 

In  the  modified  inbred  in  which  a  block  of  genes  from  teosinte  has  been 
substituted  for  a  block  of  genes  from  maize,  the  situation  is  quite  different. 
There  are  no  functional  aberrations  so  long  as  the  block  of  genes  from  teosinte 
is  heterozygous.  Under  these  circumstances  it  has  very  little  discernible 


Fig.  11.3 — Ears  of  a  teosinte-modified  inbred  strain  4R-3  which  are  isogenic  except  for  an 

introduced  block  of  genes  from  chromosome  3  of  Florida  teosinte.  The  ear  at  the  left  lacks 

the  block  of  teosinte  genes,  the  center  ear  is  heterozygous  for  it,  the  ear  at  the  right  is 

homozj'gous  for  it.  Note  the  high  degree  of  dominance  or  potence  of  the  maize  genes. 


HYBRIDIZATION   IN  THE   EVOLUTION   OF  MAIZE 


189 


effect.  Figures  11.3,  11.4,  and  11.5  show  ears  of  corn  heterozygous  for  blocks 
of  genes  from  chromosomes  3  and  4  respectively,  compared  to  "pure"  corn 
in  the  same  progeny.  The  blocks  of  genes  from  corn  are  much  more  "potent" 
(a  term  proposed  by  Wigan,  1944,  to  describe  the  integrated  dominance 
effects  of  all  genes)  than  the  block  of  genes  from  teosinte,  at  least  in  the 
striking  characteristics  which  differentiate  the  two  species.  This  is  in  itself  a 
noteworthy  phenomenon  since  corn  is  not  strongly  "dominant"  or  more  po- 
tent than  teosinte  in  the  Fi  hybrid,  where  both  species  contribute  more  or 
less  equally. 


Fig.  11.4 — These  ears  are  the  exact  counterparts"]<)f  those  in  Figure  11.3  exce])t  that  the 
block  of  teosinte  genes  was  derived  from  chromosome  4  of  Florida  teosinte. 


-< 


t 


•-^i 


\    r 


m 


Fig.  11.5 — When  the  inbred  4R-3  is  crossed  with  No.  701  the  liybrid  ear  illustrated  above 
(left)  is  produced.  When  a  modified  strain  of  4R-3  (right)  which  has  had  three  blocks  of  genes 
from  Durango  teosinte  substituted  for  corresponding  maize  genes  is  crossed  with  No.  701, 
the  hybrid  (center)  is  much  more  maize-like  than  teosinte-like.  The  hybrid,  being  multiple- 
eared,  bears  a  substantially  greater  number  of  seeds  than  either  parent  and  in  one  experi- 
ment was  appreciably  more  productive. 


HYBRIDIZATION  IN  THE  EVOLUTION  OF  MAIZE  191 

The  reason  for  the  strong  jjotence  of  maize  over  teosinte  in  blocks  of 
genes  introduced  from  teosinte  into  maize,  is  to  be  found  in  a  phenomenon 
termed  "antithetical  dominance"  which  has  been  postulated  by  Anderson 
and  Erickson  (1941)  on  theoretical  grounds.  These  writers  assumed  that  in 
species  hybrids  such  as  that  between  maize  and  Tripsacum,  the  Fi  would  be 
intermediate  but  that  backcrosses  to  either  parent  would  strongly  resemble 
the  recurrent  parent.  The  basis  for  this  assumption  is  that  the  possibilities 
for  successful  recombination  of  two  such  different  systems  is  remote. 

The  conception  of  antithetical  dominance  has  some  relationship  to 
Richey's  opinion  (1946)  that  dominance  in  some  cases  is  no  more  than  a  con- 
dition where  one  allele  is  capable  of  doing  the  entire  job,  or  most  of  it,  while 
the  other  allele  merely  stands  by.  According  to  this  interpretation,  genes  are 
not  favorable  because  they  are  dominant,  but  are  dominant  because  they  are 
favorable.  They  reveal  their  presence  by  doing  something. 

There  is,  in  any  case,  little  doubt  that  something  of  the  general  nature  of 
antithetical  dominance  or  the  kind  of  dominance  postulated  by  Richey  is 
involved  in  the  teosinte-maize  derivatives.  Both  teosinte  and  maize  are 
about  equally  potent  in  the  Fi  hybrid,  but  a  small  amount  of  teosinte  germ- 
plasm  incorporated  into  maize  in  the  heterozygous  condition  is  definitely 
lacking  in  potence. 

Effects   in   Homozygous  Condition 

Since  a  block  of  teosinte  genes  introduced  into  maize  is  largely  recessive 
in  its  effects  when  heterozygous,  its  effects  should  become  much  more  ap- 
parent in  the  homozygous  condition.  This  is  indeed  the  case.  The  ear  on  the 
right  in  Figures  11.3,  11.4,  and  11.5  illustrates  the  effects  of  one  or  more 
blocks  of  teosinte  genes  incorporated  in  a  homozygous  condition  in  the  inbred 
strain  4R-3. 

The  combination  of  traits  from  corn  and  teosinte  which  occurs  in  these 
homozygous  teosinte  derivatives  is  characterized  by  a  distinct  lack  of  har- 
mony in  the  development  of  the  pistillate  inflorescence.  The  husks  are  too 
short  for  the  ears,  the  glumes  are  too  small  for  the  kernels  and  tend  to  con- 
strict the  growing  caryopses  producing  misshapen  kernels.  The  vascular  sys- 
tem is  inadequate  for  the  number  of  kernels  borne  on  the  ear,  and  there  are 
many  shrunken  kernels  as  well  as  numerous  gaps  where  no  kernels  have  de- 
veloped. Germination  of  the  seeds  is  often  poor,  and  viability  of  short  dura- 
tion. Homozygous  combinations  of  this  kind  obviously  have  a  low  survival 
value.  Indeed  it  has  been  difficult  to  maintain  some  of  them  in  artificial 
cultures. 

These  unfavorable  effects  of  teosinte  introgression  in  the  homozygous  con- 
dition may  be  nothing  more  than  the  result  of  substituting  parts  of  one  well- 
integrated  system  for  corresponding  parts  of  another.  They  may,  however, 
also  involve  "cryptic  structural  differentiation"  of  the  kind  suggested  by 


192  PAUL  C.  MANGELSDORF 

Stephens  (1950)  for  species  of  Gossypium,  although  the  extent  of  this  cannot 
be  great,  otherwise  some  combinations  would  be  lethal.  But  whatever  the 
cause,  there  is  little  doubt  about  the  reality  of  the  unfavorable  effects. 
Therefore,  if  the  repeated  hybridization  of  corn  and  teosinte  which  has  oc- 
curred in  the  past  has  had  any  permanent  effect,  one  of  two  things  or  both 
must  have  happened:  (1)  The  undesirable  effects  of  teosinte  have  become 
recessive  as  the  result  of  natural  selection  for  modifying  factors.  (2)  The 
regions  of  the  chromatin  involving  teosinte  genes  have  been  kept  heterozy- 
gous. There  is  some  evidence  that  both  may  have  occurred. 

There  is  some  evidence,  by  no  means  conclusive,  that  maize  varieties  of 
today  have  absorbed  teosinte  germplasm  in  the  past  and  are  now  bufered 
against  the  effects  of  teosinte  genes.  There  is  at  least  no  doubt  that  when  the 
same  variety  of  teosinte  is  crossed  on  a  series  of  maize  varieties,  considerable 
variation  is  displayed  by  the  Fi  hybrids  in  the  potence  of  the  maize  parents. 

In  general,  varieties  which  show  some  evidence  of  previous  contamination 
with  teosinte  are  more  likely  to  produce  maize-like  Fi  hybrids  than  those 
which  do  not  show  evidence  of  such  contamination.  Corn  Belt  inbreds  as  a 
class  produce  the  most  maize-like  Fi  of  any  of  stocks  tested.  Figure  11.6  illus- 
trates a  case  where  a  South  American  stock  (an  inbred  strain  derived  from 
the  Guarany  corn  of  Paraguay)  is  less  potent  in  crosses  with  two  varieties  of 
teosinte  than  is  a  North  American  stock  (a  genetic  tester).  I  also  have  ob- 
served that  blocks  of  teosinte  genes  introduced  into  an  inbred  strain  of 
Guarany  by  repeated  backcrossing  have  a  greater  effect  than  these  same 
blocks  introduced  into  Texas  4R-3  or  Minn.  A158,  both  of  which  seem  al- 
ready to  contain  appreciable  amounts  of  teosinte. 

If  the  increased  potency  of  teosinte-contaminated  maize  proves  to  be  gen- 
erally true,  then  the  reason  for  it  is  that  there  has  been  a  selection  of  modify- 
ing factors  which  have  tended  to  suppress  the  most  unfavorable  conspicuous 
effects  of  the  teosinte  introgression.  Otherwise,  varieties  of  maize  containing 
teosinte  germplasm  should  produce  hybrids  which  are  more  teosinte-like, 
rather  than  more  maize-like,  than  the  average.  This  is  convincingly  demon- 
strated experimentally  by  crossing  the  original  inbred  4R-3  and  one  of  its 
modified  derivatives  with  the  same  variety  of  teosinte  (Florida  type).  The 
results  are  illustrated  in  Figure  11.7. 

The  Fi  of  4R-3  X  teosinte  is  a  typical  Fi  hybrid,  intermediate  between 
its  parents.  It  has  both  single  and  double  spikelets  and,  although  the  fact  is 
not  revealed  by  the  illustration,  it  has  approximately  the  same  type  of  re- 
sponse to  length  of  day  as  does  maize.  In  marked  contrast,  when  a  derived 
strain  of  4R-3  (in  which  a  block  of  teosinte  genes  on  chromosome  3  has  been 
substituted  for  a  corresponding  block  of  maize  genes)  is  crossed  with  the 
same  teosinte,  the  Fi  hybrid  is  scarcely  distinguishable  in  its  pistillate  spike 
from  pure  teosinte.  Furthermore,  it  has  teosinte's  response  to  length  of  day. 
Plants  of  this  hybrid  started  in  the  greenhouse  in  February  did  not  flower 


HYBRIDIZATION   IN  THE   EVOLUTION  OF  MAIZE 


193 


until  the  following  October  and  November.  This  derivative  of  a  maize-teo- 
sinte  hybrid,  therefore,  carries  at  least  two  concealed  characteristics  of  teo- 
sinte:  single  spikelets  and  response  to  length  of  day.  Genes  for  these  two  char- 
acters do  not  express  themselves  in  the  derivative  itself,  but  their  presence 
becomes  immediately  apparent  when  the  derivative  is  crossed  with  teosinte. 
The  situation  is  comparable  to  the  concealed  genes  for  hair  color  and  texture 


b^ 


i 


,  c 


% 


^ 


Fig.  11.6 — A  North  .\merican  stock  is  more  potent  in  crosses  with  Nobogame  teosinte  (A) 
and  Durango  teosinte  (C)  than  the  Guaranj'  corn  from  Paraguay  {B  and  D).  This  is  at- 
tributed to  previous  introgression  of  teosinte  accompanied  by  the  evolution  of  modifier 

complexes  in  North  American  varieties. 


in  persons  who  are  completely  bald.  The  genes  are  there  but  have  no  oppor- 
tunity to  express  themselves. 

Since  varieties  of  maize  which  appear  to  be  the  product  of  previous  teo- 
sinte contamination,  such  as  those  of  the  Corn  Belt,  behave  quite  differently 
in  crosses  from  stocks  known  to  be  contaminated,  there  is  at  least  an  indica- 
tion that  such  contamination  has  become  modified  through  selection  acting 
upon  the  modifier  complex.  More  data  are  obviously  needed  on  this  problem. 

A  second  question  which  arises  in  considering  the  effects  of  the  natural 
hybridization  of  com  and  teosinte  is  whether  there  is  any  mechanism  which 


194 


PAUL  C.  MANGELSDORF 


tends  to  maintain  the  maize-teosinte  loci  in  a  perpetual  state  of  heterozy- 
gosity. It  already  has  been  shown  that  homozygous  teosinte  genes  in  the 
maize  complex  are  decidedly  deleterious.  Therefore,  if  the  teosinte  genes  are 


^*"«*W*^iW- 


D 


Fig.  11.7 — When  the  inbred  4R-3(^)  iscrossed  with  Florida  teosinte  (C),  thcFi  hybrid  ears 
(B)  are  maize-like  in  having  four-ranked  ears,  some  double  spikelets,  and  partially  naked 
seeds.  When  a  teosinte-modified  strain  of  4R-3  (D)  is  crossed  with  Florida  teosinte  (F), 
the  Fi  hybrid  (E)  is  much  more  teosinte-like.  The  spikes  are  two-ranked,  single,  and  the 
seeds  are  completely  enclosed.  The  teosinte  derivative  obviously  carries  "concealed"  genes 

for  these  teosinte  characteristics. 


to  survive  their  deleterious  effects,  they  must  be  modified  through  selection 
or  the  genes  must  be  maintained  in  a  more  or  less  heterozygous  state.  It  may 
be  assumed  that  the  latter  mechanism  would  operate  only  if  heterozygosity 
for  a  group  of  maize-teosinte  genes  confers  a  distinct  selective  advantage 
making  the  heterozygous  combination  superior,  not  only  to  the  homozygous 
teosinte  genes  (as  it  obviously  is)  but  also  to  the  corresponding  homozygous 
maize  genes. 


HYBRIDIZATION  IN  THE  EVOLUTION  OF  MAIZE  195 

Data  are  available  both  from  my  experiments  and  those  of  R.  G.  Reeves 
(1950),  conducted  independently,  to  indicate  that  heterozygosity  for  a  block 
of  teosinte  genes  does  sometimes  confer  a  selective  advantage.  In  1944,  in  my 
experiments,  five  Corn  Belt  inbred  strains  were  crossed  with  the  Texas  in- 
bred 4R-3,  as  well  as  with  foar  modified  strains  of  4R-3  in  which  teosinte 
genes  had  been  substituted  for  maize  genes.  The  four  modified  strains  may  be 
briefly  described  as  follows: 

No.   Blocks  Teosinte 

Strain  Genes  Variety 

Modified  4R-3  Strain  A 2  Florida 

Modified  4R-3  Strain  B 2  Florida 

Modified  4R-3  Strain  C 3  Durango 

Modified  4R-3  Strain  D 3  Durango 

The  Fi  hybrids  were  grown  in  1945  in  two  replications  in  a  modified  Latin- 
Square  yield  test.  Several  hybrids  were  omitted  for  lack  of  sufficient  seed. 
The  results  are  shown  in  Table  11.4. 

TABLE  11.4 

AVERAGE   YIELDS   IN   BUSHELS  PER   ACRE  OF  HYBRIDS  OF 

CORN  BELT  INBREDS  WITH  TEXAS  4R-3  AND  ITS 

TEOSINTE-MODIFIED  DERIVATIVES 


4R-3  OR  Deriv.^tive 

Corn  Belt  Inbreds 

K1S5 

38-11 

L317 

701 

CC24 

4R-3  (check) 

108.6 

102.6 

126.6* 

94.2 

93.0 

85.2 

99.0 

100.2 
87.0 

100  2 

Modified  Strain  A 

88  8 

Modified  Strain  B 

82.8 
75.6 
57.0 

109.8 
66.0 
71.4 

78  6 

Modified  Strain  C 

Modified  Strain  D 

97.8 
146.4* 

92.4 
79.8 

*  Difference  probably  significant. 


Of  the  17  hybrids  tested,  only  3  proved  to  be  better  than  the  correspond- 
ing checks  in  total  yield,  and  in  only  2  of  these  is  the  difference  significant. 
Although  the  data  are  not  extensive,  there  is  some  indication  that  the  Corn 
Belt  inbred  strains  used  in  these  experiments  differ  in  their  ability  to  "com- 
bine" with  the  teosinte  derivatives. 

Perhaps  more  important  than  total  yield,  from  the  standpoint  of  selective 
reproductive  advantage,  is  total  number  of  seeds  per  plant  (Table  11.5). 
Here  6  of  the  15  hybrids  for  which  data  are  available  were  superior  to  the 
checks,  4  of  these  significantly  so. 

These  results,  so  far  as  they  go,  are  in  agreement  with  the  recently  pub- 
lished results  of  Reeves  (1950).  Reeves  tested  49  modified  4R-3  lines  in  hy- 
brids with  a  common  tester.  He  found  none  significantly  better  than  the 
check  in  yield,  although  several  were  superior  in  heat-tolerance.  Reeves 


196 


PAUL  C.  MANGELSDORF 


found,  however,  that  when  teosinte  germplasm  was  introduced  into  another 
inbred  strain,  127C,  the  results  obtained  in  the  hybrids  were  somewhat  dif- 
ferent. In  1946,  6  hybrids  out  of  25  were  better  than  the  check,  3  of  them 
significantly  so.  In  1947,  15  hybrids  out  of  49  were  better  than  the  check, 
6  of  them  significantly  so.  Reeves  suggested  that  the  difference  between  4R-3 
and  127C  in  their  response  to  teosinte  introgression  lies  in  the  fact  that 
4R-3  already  contained  considerable  amounts  of  teosinte  germplasm  while 
127C  does  not.  The  suggestion  is  supported  by  differences  in  the  morphologi- 
cal characteristics  of  the  two  lines. 

There  was  also  an  indication  in  Reeves'  experiments  that  the  entries  with 

TABLE  11.5 

AVERAGE  NUMBERS  OF  SEEDS  PER  PLANT  IN  HYBRIDS  OF 

CORN  BELT  INBREDS  WITH  TEXAS  4R-3  AND  ITS 

TEOSINTE-MODIFIED  DERIVATIVES 


Corn  Belt  Inbreds 

4R-3  OR  Derivative 

K1S5 

38-11 

L317 

701 

CC24 

4R-3  (check)                  .... 

849 
756 
937 
1419* 
770 

636 

925 

1132 
1095 

1179 

Modified  Strain  A 

807 

Modified  Strain  B 

809' 
573 

1107* 
746 
843 

Modified  Strain  C 

Modified  Strain  D 

1696* 
1811* 

885 
864 

*  Difference  probably  significant. 


teosinte  genes  made  their  best  showing  in  1947,  a  season  of  severe  drought. 

Considering  all  of  the  results  together  it  may  be  concluded  that:  (c) 
blocks  of  teosinte  genes  in  the  heterozygous  condition  do  in  some  instances 
improve  the  total  yield  of  the  plants  which  contain  them ;  (b)  even  more  fre- 
quently do  such  blocks  of  genes  increase  the  total  number  of  seeds  produced ; 
(c)  there  is  some  evidence  that  the  teosinte  derivatives  impart  resistance  to 
heat  and  drought  to  their  hybrids. 

In  those  crosses  in  which  the  heterozygous  combination  is  superior  to 
either  of  the  homozygous  combinations,  a  block  of  maize  genes  or  a  block  of 
teosinte  genes,  natural  selection  would  undoubtedly  favor,  at  least  initially, 
the  heterozygous  combination.  If  the  block  of  genes  were  one  involving  the 
region  of  the  centromere  where  crossing-over  is  reduced,  it  is  quite  possible 
that  the  block  of  genes  would  be  retained  more  or  less  intact  for  a  consider- 
able number  of  generations.  The  maintenance  of  heterozygosity  through 
natural  selection  also  would  be  promoted  if,  as  in  the  case  of  Drosophila 
studied  by  Dobzhansky,  one  set  of  genes  is  superior  in  adapting  the  organism 
to  one  kind  of  environment  while  the  other  set  contributes  to  adaptation 


HYBRIDIZATION   IN  THE   EVOLUTION  OF  MAIZE  197 

to  a  wholly  different  environment  which  the  organism  also  encounters  pe- 
riodically. 

It  cannot  be  proved  that  such  a  situation  exists  in  the  case  of  maize  which 
has  become  contaminated  with  teosinte,  but  it  is  quite  possible  that  it  does. 
For  example,  human  selection  when  practiced  would  tend  to  favor  the  larger- 
seeded,  larger-eared  individuals  with  a  minimum  of  teosinte  contamination. 
Natural  selection  would  favor  the  individuals  with  the  larger  number  of 
seeds,  hence  those  with  an  appreciable  amount  of  teosinte  contamination. 
These  two  forces  operating  simultaneously  or  alternately  would  tend  to  per- 
petuate the  heterozygote.  Similarly,  if  maize  germplasm  were  superior  in 
seasons  of  excessive  moisture  and  teosinte  germplasm  in  seasons  of  drought 
(for  which  there  is  some  evidence),  there  would  be  a  tendency  for  natural 
selection  to  perpetuate  heterozygous  combinations.  It  cannot  be  demonstrated 
that  any  of  these  hypothetical  situations  actually  exist.  There  is  no  doubt, 
however,  that  present-day  maize  is  highly  heterozygous,  and  there  is  more 
than  a  suspicion  that  repeated  hybridization  with  teosinte  has  been  respon- 
sible for  part  of  the  heterozygosity. 

DISCUSSION 

The  present-day  heterozygosity  of  maize  may  involve  a  variety  of  differ- 
ent factors  and  forces  which  have  operated  during  its  past  history.  Two  of 
these  are  now  reasonably  clear:  interracial  hybridization,  and  introgression 
of  teosinte  into  maize. 

When  interracial  hybridization  occurs,  hybrid  vigor  not  only  manifests 
itself  in  the  first  generation,  but  also  persists  in  part  through  an  indefinite 
number  of  subsequent  generations.  Maize  under  domestication  is,  therefore, 
potentially  a  self-improving  plant.  The  evidence  from  Mexican  races  of 
maize  indicates  that  repeated  interracial  hybridization  has  been  an  extremely 
important  factor  in  the  evolution  of  maize  in  Mexico.  There  is  every  reason 
to  believe  that  the  situation  in  Mexico,  so  far  as  interracial  hybridization  is 
concerned,  is  typical  of  other  parts  of  America. 

The  second  factor,  introgression  of  teosinte,  which  is  believed  to  have 
played  an  important  role  in  the  evolution  of  maize,  is  not  so  easily  demon- 
strated. There  is  no  doubt,  however,  that  teosinte  is  hybridizing  with  maize 
in  Guatemala  and  Mexico  today,  or  that  this  hybridization  has  occurred  in 
the  past.  It  would  be  surprising  indeed  if  such  hybridization  had  no  effect 
upon  the  evolution  of  maize.  There  is  every  indication  that  it  has  had  a  pro- 
found effect.  All  of  the  most  productive  modern  agricultural  races  of  maize 
in  Mexico  show  evidence  of  contamination  with  teosinte,  not  only  in  their 
external  characters,  but  also  in  their  internal  cytological  characteristics. 

It  can  be  shown  experimentally  that  teosinte  germplasm,  when  introduced 
into  maize,  may  sometimes  have  a  beneficial  effect  when  heterozygous,  but 
is  always  deleterious  when  homozygous.  Therefore  it  follows  that  after  maize 


198  PAUL  C.  MANGELSDORF 

and  teosinte  have  hybridized,  and  after  there  has  been  an  introgression  of 
teosinte  into  maize:  (1)  the  teosinte  genes  must  be  eliminated  or,  (2)  their 
effects  must  be  changed  through  the  accumulation  of  a  new  modifier  com- 
plex, or  (3)  they  must  be  kept  in  a  heterozygous  state.  There  is  evidence,  but 
not  final  proof,  that  both  of  the  two  last-named  factors  have  operated  during 
the  evolution  of  maize.  Interracial  and  interspecific  hybridization  accom- 
panied by  sustained  heterosis  are  therefore  regarded  as  two  important  fac- 
tors in  the  evolution  of  maize. 

SUMMARY 

1.  Evidence  is  presented  to  show  that  both  interracial  and  interspecific 
hybridization,  accompanied  by  heterosis,  have  been  factors  in  the  evolution 
of  maize. 

2.  The  races  of  maize  of  Mexico  are  cited  as  an  example  of  interracial  hy- 
bridization. Of  the  25  Mexican  races  described  by  Wellhausen  et  al.,  14  are 
considered  to  be  the  products  of  interracial  hybridization. 

3.  The  hybrid  vigor,  which  occurs  when  races  of  maize  are  crossed,  is 
capable  of  persisting  in  part  in  subsequent  generations.  Maize  under  domesti- 
cation is  therefore  potentially  a  self-improving  plant. 

4.  Interspecific  hybridization  of  maize  and  teosinte  is  occurring  in  Gua- 
temala and  Mexico  today,  and  there  is  evidence — archaeological,  morphologi- 
cal, and  cytological — that  it  has  occurred  in  the  past. 

5.  Introgression  of  teosinte  into  maize  in  experimental  cultures  is  some- 
times beneficial  when  the  teosinte  genes  are  heterozygous,  but  is  always 
deleterious  when  they  are  homozygous. 

6.  It,  therefore,  seems  probable  that  the  persistence  of  teosinte  germ- 
plasm  in  races  of  maize  has  been  accompanied  either  by  development  of 
modifier  complexes  which  have  made  the  teosinte  genes  recessive  in  their 
action,  or  by  the  maintenance  of  a  continued  state  of  heterozygosity. 

7.  The  possibility  that  heterozygosity  in  maize  has  been  preserved  by 
natural  selection  as  it  has  been  in  Drosophila  is  discussed. 


STERLING  EMERSON 

California  Insfifufe  of  Technology 


Chapter   12 

Biochemicol  Models 
of  Heterosis  in  Neurosporo 


Some  of  the  things  that  have  been  learned  about  gene  controlled  reactions 
in  Neurospora  can  be  used  in  forming  a  picture  of  how  individual  genes  con- 
tribute to  heterosis.  I  wish  to  consider  especially  those  examples  which  indi- 
cate that  heterozygosity  at  a  single  locus  may  influence  the  growth  of  an 
organism  to  a  considerable  extent. 

It  should  be  noted  at  the  beginning,  however,  that  one  is  not  justified  in 
assuming  that  the  situations  found  in  Neurospora  are  necessarily  similar  to 
those  occurring  in  the  higher  organisms  in  which  heterosis  is  ordinarily 
studied.  It  may  be  unwise  to  assume  that  any  two  organisms  are  essentially 
similar.  There  are  special  reasons  for  caution  in  making  comparisons  between 
Neurospora  and  higher  plants  and  animals,  since  the  nuclear  and  chromo- 
somal basis  for  the  expression  of  heterosis  is  so  dissimilar.  On  the  other  hand, 
there  is  a  considerable  accumulation  of  information  about  the  parts  played 
in  the  physiology  and  biochemistry  of  Neurospora  by  individual  genes 
(Beadle,  1948;  Horowitz,  1950)  and,  with  proper  caution,  we  may  assume 
that  some  of  this  information  may  have  rather  broad  application. 

In  any  haploid  organism,  such  as  the  ascomycetous  fungus  Neurospora, 
in  which  there  is  a  single  set  of  genes  in  each  nucleus,  such  phenomena  as 
dominance,  heterozygosis,  and  heterosis  cannot  occur.  There  is,  however, 
a  condition  known  as  heterocaryosis  which  permits  a  loose  approximation 
to  each. 

CHARACTERISTICS  OF  HETEROCARYONS 

The  plant  body  of  Neurospora  can  be  said  to  be  made  up  of  cells,  but  they 
are  very  different  from  the  cells  of  higher  plants.  In  the  first  place,  the  cells 
contain  a  large  and  variable  number  of  nuclei  in  a  common  cytoplasm.  The 
so-called  cells  themselves  are  not  as  discrete  as  cells  are  generally  supposed 

199 


200 


STERLING  EMERSON 


to  be.  The  walls  between  them  have  perforations  which  permit  both  cyto- 
plasm and  nuclei  to  move  from  cell  to  cell.  If  all  nuclei  are  identical,  their 
movement  and  distribution  is  probably  of  minor  importance,  but  if  they  are 
not  identical  there  may  be  effects  of  considerable  consequence  arising  from 
irregularities  in  nuclear  distribution. 

There  are  two  ways  in  which  a  mixture  of  different  kinds  of  nuclei  within 
a  single  cell  may  come  about.  In  the  growth  resulting  from  a  sexually  pro- 
duced ascospore,  or  from  a  uninucleate  asexual  microconidium,  all  nuclei 
are  directly  descended  from  a  single  haploid  nucleus.  Barring  mutation,  they 
should  all  have  the  same  genetic  constitution.  After  the  growth  has  become 


Strain  X 


Fig.  12.1 — Heterocaryon  formation  resulting  from  hyphal  fusion  (a  diagram). 


multinucleate,  if  a  mutation  should  occur  in  one  nucleus,  the  descendants  of 
that  nucleus  would  then  have  a  different  genetic  constitution  from  the  re- 
maining nuclei  in  the  common  cytoplasm,  and  a  condition  of  heterocaryosis 
would  exist.  The  second  way  in  which  heterocaryons  arise  is  from  the  direct 
fusion  of  branches  or  hyphae  of  different  strains,  with  the  subsequent  in- 
termingling of  their  nuclei.  By  the  latter  method,  heterocaryons  of  pre- 
determined genetic  constitution  can  be  made  at  will. 

The  controlled  production  of  heterocaryons  is  shown  diagrammatically  in 
Figure  12.1.  Strain  X  is  represented  as  having  black  nuclei  to  distinguish 
them  from  the  nuclei  of  strain  Y,  which  are  pictured  as  being  white.  After 
fusion  between  hyphae,  nuclei  of  strain  Y  may  migrate  into  cells  of  strain  X, 
and  those  of  X  into  Y.  It  is  possible  that  different  hyphal  tips,  growing  from 
this  common  mass  of  cells,  will  have  different  relative  numbers  of  the  two 
sorts  of  nuclei,  as  illustrated  by  the  ratios  1:7,  1:1,  and   7:1  in  three  of 


BIOCHEMICAL  MODELS  OF   HETEROSIS  IN   NEUROSPORA 


201 


the  hy})hal  tips.  To  prove  that  two  kinds  of  nuclei  were  present  in  the  same 
cells  of  such  heterocaryons,  Beadle  and  Coonradt  (1944)  cut  ofT  single  hyphal 
tips,  transferred  them  to  fresh  medium,  and  then  identified  two  kinds  of 
nuclei  in  the  resulting  growth  by  genetic  test. 

Where  there  is  freely  branching  filamentous  growth,  as  in  Neurospora,  it  is 
possible  for  the  two  types  of  nuclei  in  a  heterocaryon  to  become  sorted  out 


Fig.  12.2 — Somatic  segregation  of  dissimilar  nuclei  in  the  formation  of  conidia  (a  diagram). 


purely  as  a  matter  of  chance,  as  illustrated  in  a  schematic  way  in  Figure  12.2. 
This  diagram  actually  represents  an  erect  fruiting  branch,  or  conidiophore, 
on  which  the  asexual  spores  are  born.  The  conidia  of  Neurospora  have 
variable  numbers  of  nuclei,  but  generally  more  than  one.  Dodge  (1942) 
proved  that  two  kinds  of  nuclei  were  present  in  the  same  cell  of  a  heterocar- 
yon by  growing  cultures  from  single  conidia,  and  then  showing  by  genetic 
test  that  some  of  these  cultures  had  both  types  of  nuclei.  In  some  instances 
he  was  able  to  distinguish  the  heterocaryotic  and  both  homocaryotic  types 
in  culture  derived  from  single  conidia  by  their  mon:)hological  characteristics. 
The  essential  differences  between  Neurospora  and  higher  organisms  with 


202  STERLING  EMERSON 

respect  to  heterosis  result  from  the  points  just  noted.  In  a  diploid  which  is 
heterozygous  for  a  single  gene  pair,  both  alleles  are  present  in  the  same  nu- 
cleus and  in  equal  dosage.  Whereas  in  the  corresponding  haploid  heterocar- 
yon,  the  two  alleles  are  present  in  different  nuclei,  and  the  relative  propor- 
tions of  the  two  alleles  vary  with  the  frequencies  of  the  two  types  of  nuclei. 
All  cells  of  a  diploid  heterozygote  have  the  same  genetic  constitution,  but 
there  can  be  a  considerable  variation  in  genetic  constitution  in  different  parts 
of  a  heterocaryotic  individual.  Interactions  between  alleles,  by  which  I  mean 
such  things  as  the  expression  of  dominance,  must  result  from  the  ability  of 
genes  to  act  at  some  distance  in  heterocaryons,  in  which  there  is  no  possibility 
of  an  intimate  association  of  alleles  within  a  nucleus  (Lewis,  1950).  It  is 
considerations  such  as  these  that  show  that  dominance  and  heterosis-like 
effects  in  Neurospora  are  only  approximations  to  the  phenomena  as  known  in 
diploid  organisms. 

HETEROSIS  IN  HETEROCARYONS 

An  enhancement  of  growth,  closely  simulating  heterosis,  in  heterocaryons 
of  Neurospora  teirasperma  was  reported  by  Dodge  in  1942.  In  this  paper  he 
distinguished  between  heterocaryotic  vigor  and  the  hybrid  vigor  of  diploid 
organisms  along  much  the  same  lines  as  I  have  just  done.  He  suggested  that 
the  heterocaryotic  vigor  observed  might  be  the  result  of  complementing 
growth  factors  whose  production  was  controlled  by  the  two  types  of  nuclei 
(Robbins,  1950).  It  was  later  (Dodge,  Schmitt,  and  Appel,  1945)  demon- 
strated that  genes  responsible  for  enhanced  growth  segregated  and  recom- 
bined  in  a  normal  fashion.  These  studies  showed  that  genes  residing  in  differ- 
ent nuclei,  but  in  a  common  cytoplasm,  can  cooperate  in  establishing  condi- 
tions favoring  rapid  growth,  and  that  a  condition  resembling  hybrid  vigor 
occurs. 

Meantime,  Beadle  and  Coonradt  (1944)  had  reported  on  heterocaryons 
between  pairs  of  mutant  strains  of  Neurospora  crassa,  each  of  which  is  unable 
to  synthesize  a  particular  vitamin  or  amino  acid.  Each  mutant  strain  by  itself 
is  unable  to  grow  unless  supplied  with  its  specific  growth  requirement,  but 
nine  heterocaryons  involving  different  combinations  of  seven  mutant  strains 
grew  at  rates  approximating  that  of  wild  type  without  the  addition  of  growth 
factors.  The  authors  conclude  that  the  wild  type  allele  is  dominant  to  the 
mutant  allele  in  each  of  the  examples  studied. 

Beadle  and  Coonradt  note  further  that  in  such  heterocaryons,  in  which 
there  is  the  opportunity  for  great  diversity  in  the  relative  numbers  of  the 
two  types  of  nuclei  in  different  hyphal  tips,  those  tips  having  the  most  favor- 
able proportions  of  nuclei  should  grow  most  rapidly.  Conversely,  rapidly 
growing  hyphae  should  have  the  two  sorts  of  nuclei  in  roughly  optimal  pro- 
portions. In  heterocaryons  involving  pairs  of  mutant  strains,  Beadle  and 
Coonradt  found  nuclear  ratios  varying  between  approximately  1 : 1  and  al- 
most 20: 1.  They  interpreted  these  results  to  mean  that  the  wild  type  alleles 


BIOCHEMICAL  MODELS  OF  HETEROSIS  IN  NEUROSPORA  203 

of  different  mutant  genes  have  different  degrees  of  dominance.  A  strongly 
dominant  wild  type  allele  will  need  to  be  present  in  relatively  few  nuclei — say 
one  in  twenty. 

A  heterocaryon  between  two  mutant  strains  could  grow  at  the  maximum 
rate  over  a  large  range  of  nuclear  proportions,  provided  the  wild  type  alleles 
concerned  were  both  strongly  dominant.  A  weakly  dominant  wild  type 
allele,  on  the  other  hand,  must  be  present  in  a  large  j)roportion  of  the  nuclei — 
say  nineteen  of  twenty — to  ensure  vigorous  growth.  Heterocaryons  in  which 
the  wild  type  alleles  concerned  are  both  weakly  dominant  could  never  result 
in  vigorous  growth,  since  the  two  wild  type  alleles  cannot  both  be  present 
in  excess,  one  being  in  one  type  of  nucleus  and  the  other  in  the  remaining 
nuclei. 

HETEROSIS  DUE  TO  HETEROZYGOSITY  AT  ONE  LOCUS 

The  heterosis  effect  in  heterocaryons  studied  by  Beadle  and  Coonradt  re- 
sults from  the  mutually  complementary  nature  of  the  nuclei  involved.  For 
each  deleterious  mutant  allele  in  one  nucleus  there  is  the  corresponding 
favorable  and  dominant  wild  type  allele  in  another.  In  contrast  to  these 
there  are  other  heterocaryons  (briefly  reported  in  Emerson,  1947)  in  which 
the  nuclei  differ  in  only  one  gene,  yet  which  still  show  the  heterosis  effect. 
Heterocaryons  in  which  some  nuclei  carry  the  dominant  allele  and  some  the 
recessive  are  superior  to  homocaryons,  all  of  whose  nuclei  have  the  dominant 
allele,  or  all  the  recessive. 

Heterocaryotic  Suppression  of  the  Sulfonamide-requiring  Character 

Most  of  the  heterocaryons  of  this  sort  that  have  been  found  so  far  have 
involved  the  so-called  sulfonamide-requiring  mutant  strain.  At  35°  on  mini- 
mal medium,  this  strain  makes  extremely  poor  growth,  but  it  does  keep 
creeping  along.  After  varying  lengths  of  time,  it  frequently  happens  that  the 
growth  will  change  to  a  rapid  vigorous  type.  Growth  curves  of  six  cultures 
which  have  reverted  to  something  approaching  wild  type  growth  are  shown 
in  Figure  12.3.  When  the  mycelium  had  reached  the  end  of  the  growth  tubes, 
inocula  from  the  newest  growth  were  introduced  into  fresh  tubes  containing 
minimal  medium,  resulting  in  the  growth  curves  shown  in  the  upper  part  of 
the  figure. 

From  these  curves  it  can  be  seen  that  the  reverted  type  of  growth  usually 
persists  through  a  conidial  transfer.  After  the  mycelium  had  reached  the  end 
of  the  second  tube,  conidia  were  removed  and  used  in  outcrosses  to  wild  type 
to  determine  the  genetic  constitution  of  their  nuclei.  These  tests  showed 
that  each  of  the  six  cultures  represented  in  Figure  12.3  was  a  heterocaryon. 
One  type  of  nucleus  present  in  each  heterocaryon  was  identical  to  those  in 
the  original  sulfonamide-requiring  strain.  The  second  type  of  nucleus  in  each 
also  carried  the  sulfonamide-requiring  gene,  sjo  (in  one  instance,  that  de- 
rived from  culture  number  1  in  Figure  12.3,  the  sJo  gene  itself  was  somewhat 


en 


bo  On 


>.  ni 

^  S 

^  eg 

O  n! 
u 

-2  Pi 

-a  o 


.S  -^ 

o  - 

OS    f 

CI-  « 

c/i    C 

2  ^ 

3    S 

o  3 

c  ^ 

3     M 

cr  u 

11    o 


en 

cj  .5 

c  — 

'^   o 


> 


O 


O 

o 

c 


3 

O 

o 


BIOCHEMICAL  MODELS  OF   HETEROSIS  IN   NEUROSPORA 


205 


modified),  and  in  addition  a  second  mutant  gene,  S,  which  was  presumably 
responsible  for  the  change  in  growth  (Table  12.1). 

The  new  mutants  appearing  in  the  heterocaryons  have  been  called  sup- 
pressors because  they  overcome  the  deleterious  effect  of  the  sulfonamide- 
requiring  gene  in  heterocaryons.  Actually  they  are  not  like  the  usual  suj)- 
pressors,  because  in  homocaryotic  strains  which  also  carry  the  sulfonamide- 
requiring  gene  they  do  not  result  in  wild  type  growth. 

Growth  characteristics  of  strains  homocaryotic  for  four  of  these  suppres- 
sors, with  and  without  the  sulfonamide-requiring  gene,  are  represented  in 


TABLE  12.1 

DISTRIBUTION     OF     NUCLEI     IN 
THE  HETEROCARYONS  REPRE- 
SENTED IN  FIGURE  12.3 


From  Culture 
Tube  Number 

Nuclei 

sfo. 

+ 

sfo,S 

1. 

2. 
3. 
4. 

3 

6 

15 

8 
8 
1 

5 
5 
2 
1 

5 

1 

6 

14 

Figure  12.4.  From  these  growth  curves  it  can  be  seen  that  wild  type  (+,  +) 
is  neither  inhibited  by  sulfanilamide  in  a  concentration  of  2  X  10~^  M,  nor 
stimulated  by  />-aminobenzoic  acid  in  a  concentration  of  10"'*  M  when  grown 
at  35°,  and  is  only  slightly  inhibited  by  sulfanilamide  at  25°.  At  35°  growth  of 
the  sulfonamide-requiring  strain  (sfo,  -\-)  is  stimulated  by  sulfanilamide  and 
inhibited  by  ^-aminobenzoic  acid,  though  neither  substance  has  an  appre- 
ciable effect  at  25°  in  the  concentrations  used. 

The  suppressor  from  tube  1  (+,  ^-1)  does  not  grow  at  35°,  and  grows  slow- 
ly on  all  media  at  25°.  The  suppressor  from  tube  2  (+,  5-2)  differs  from  wild 
type  principally  in  taking  longer  to  attain  its  maximum  growth  rate,  though 
there  is  also  some  stimulation  by  sulfanilamide  at  35°.  When  combined  as  a 
double  mutant  with  the  sulfonamide-requiring  gene  (sfo,  5-2),  it  almost  ap- 
proximates the  growth  of  wild  type.  The  suppressor  from  tube  4  (+,  5-4) 
differs  from  wild  type  in  being  stimulated  by  />-aminobenzoic  acid  and  in- 
hibited by  sulfanilamide,  the  inhibition  being  stronger  at  25°.  In  combination 
with  the  sulfonamide-requiring  gene  (sfojSA)  it  resembles  the  sulfonamide- 
requiring  strain  itself  except  that  there  is  a  long  lag  phase  on  sulfanilamide 
at  35°,  and  inhibition  at  25°.  The  suppressor  from  tube  6,  either  alone 


HOURS 

Fig.  12.4 — Growth  curves  of  suppressor  strains  in  absence  and  presence  of  the  sulfonamide-requiring  gene  (sfo) 
at  35°  and  25°  on  minimal  medium  (light  Hne),  10"*  M  />-aminobenzoic  acid  (line  of  open  circles  —  PABA),  and 

2  X  10""  M  sulfanilamide  (dotted  line  -SA). 


BIOCHEMICAL  MODELS  OF  HETEROSIS  IN  NEUROSPORA  207 

(-|_^  5.6)  or  in  combination  with  the  sulfonamide-requiring  gene  (sfo,  5-6), 
grows  very  poorly  at  35°. 

Of  those  illustrated,  suppressors  numbered  4  and  6  are  perhaps  the  most 
significant  to  the  present  discussion.  When  combined  with  the  sulfonamide- 
requiring  gene  (sfo,  SA  and  sfo,  S-6),  neither  grows  well  on  minimal  medium 
at  35°.  Yet  heterocaryons  between  either  of  these  double  mutants  and  the 
sulfonamide-requiring  strain  are  enabled  to  grow  quite  well  under  those 
conditions.  In  these  heterocaryons  the  sulfonamide-requiring  gene  is  present 
in  all  nuclei,  in  some  of  which  it  is  combined  with  a  suppressor.  The  suppres- 
sor is  not  capable  of  overcoming  the  ill  effects  of  the  sulfonamide-requiring 
gene  when  present  in  all  nuclei,  but  is  effective  when  present  in  only  some 

of  them. 

Biochemical    Basis   for   the   Sulfonamide-requiring    Character 

This  seeming  paradox  becomes  less  important  once  the  nature  of  the  reac- 
tion controlled  by  the  sulfonamide-requiring  gene  is  understood  (Zalokar, 
1948,  1950;  Emerson,  1950).  The  diagrams  in  Figure  12.5  illustrate  some  of 
the  important  reactions  involved.  There  are  a  large  number  of  amino  acids, 
vitamins,  components  of  nucleic  acid,  and  so  on,  that  are  essential  to  growth. 
But  we  shall  consider  only  two  amino  acids,  methionine  and  threonine,  and 
the  vitamin  />-aminobenzoic  acid.  Para-aminobenzoic  acid  is  involved  in  a 
number  of  reactions  essential  to  growth,  one  of  which  is  the  final  step  in  the 
synthesis  of  methionine  from  homocysteine.  Wild  type  carries  out  all  essen- 
tial reactions  and  produces  all  essential  growth  factors,  with  the  exception 
of  biotin  which  must  be  supplied  to  all  strains. 

The  reaction  governed  by  the  sulfonamide-requiring  gene  has  not  yet  been 
identified,  but  we  know  quite  a  little  about  it.  It  requires  the  presence  of  both 
homocysteine  and  />-aminobenzoic  acid.  Presumably  homocysteine  is  used  as 
a  substrate  in  this  reaction,  and  />-aminobenzoic  acid,  or  a  derivative,  is 
needed  as  a  catalyst.  The  reaction  either  results  in  the  destruction  of  threo- 
nine or  else  interferes  with  its  normal  utilization,  so  that  the  sulfonamide- 
requiring  strain  has  too  little  threonine  for  growth.  W^e  also  know  that  more 
homocysteine  is  required  for  this  deleterious  reaction  than  for  the  syn- 
thesis of  methionine,  and  that  in  the  presence  of  limiting  amounts  of  homo- 
cysteine, the  synthesis  of  methionine  goes  on  without  any  interference  with 
the  utilization  of  threonine. 

Furthermore,  the  deleterious  reaction  requires  larger  amounts  of  ^-amino- 
benzoic  acid  than  are  needed  for  all  essential  reactions  combined.  Only  about 
half  as  much  is  needed  in  the  synthesis  of  methionine,  about  a  quarter  as 
much  in  the  production  of  purines,  and  very  much  less  still  for  other  essen- 
tial, but  still  unidentified  factors.  Both  wild  type  and  the  sulfonamide-requir- 
ing strain  produce  about  one  hundred  times  as  much  />-aminobenzoic  acid  as 
is  needed  for  all  essential  reactions. 

We  know  of  three  ways  in  which  the  deleterious  reaction  leading  to  threo- 


Wild  Type 


Homocysteine 

CH2-SH 

CH2 
"^  CH-NHz 

COOH 


Methionine 
CH2-S-CH3 

CH2 

CH-NH2 

COOH 


COOH 

^-Aminobenzoic 
acid 


GROWTH 


CH3 
^CHOH 

CH-NH2 
COOH 

Threonine 


Sulfonamide  Requiring,  Homocvsteineless 


H-9a 


*-  .»-.■■- 


•-  c?/f(yr  ^ 


Fig.  12.5— 
by  genes  of 


CH3 

CHOH 

CH-NH2 

COOH 


Certain  biochemical  reactions  essential  to  growth  in  Neurospora  as  influenced 
the  sulfonamide-requiring  strain  (E-15172),  the  homocysteineless  strain  (H-98), 
and  the  aminobenzoicless  strain  (1633). 


208 


SulfonoTiide  Reg uirin g 


CH2-SH  , 

CH2    ^'''' 
CH-NHz 
COOH    "^2 


CH2-S-CH3 
GHz 

CH-NHg 
COOH 


UNITS 


UNITS^     UNITS 


COOH 


1/2  UNIT 


CH3 

CHOH 

CH-NHz 

COOH 


GROWTH 


Sulfonamide  Requiring,  Aminobenzoicless 


CH^-S-CH3 


CH2-SH 
CH2 
CH-NH2 
COOH 


CH3 
^CHOH 

"^'cH-NHz 
COOH 


Fig.  12.5 — Continued 


GROWTH 


209 


210  STERLING  EMERSON 

nine  deficiency  can  be  prevented  by  genetic  means.  The  simplest  is  of  course 
by  introducing  the  wild  type  allele  of  the  sulfonamide-requiring  gene,  but  the 
other  two  are  of  more  interest.  One  of  these  is  by  introducing  a  genetic  block 
to  the  synthesis  of  homocysteine.  Mutant  strain  H-98  blocks  the  terminal 
step  in  the  synthesis  of  homocysteine.  In  the  double  mutant — sulfonamide- 
requiring,  homocysteineless — there  is  no  interference  with  the  availability  of 
threonine  for  growth,  since  the  deleterious  reaction  does  not  take  place  in 
the  absence  of  homocysteine.  In  the  absence  of  homocysteine,  however,  there 
can  be  no  synthesis  of  methionine,  so  that  the  double  mutant  fails  to  grow 
because  of  a  methionine  deficiency.  The  double  mutant  will  grow  if  supplied 
with  exactly  the  right  amount  of  methionine — more  inhibits  growth,  because 
methionine  is  degraded  to  homocysteine  which  then  supports  the  deleterious 
reaction  (Zalokar,  1950). 

The  remaining  method  is  to  introduce  a  genetic  block  to  the  synthesis  of 
/»-aminobenzoic  acid.  In  the  double  mutant — sulfonamide-requiring,  amino- 
benzoicless — there  is  again  no  interference  with  the  utilization  of  threonine 
since  there  is  no  />-aminobenzoic  acid  to  catalyse  the  deleterious  reaction. 
There  is  again  a  deficiency  for  methionine,  because  />-aminobenzoic  acid  is 
needed  in  its  synthesis.  There  is  also  a  deficiency  of  />-aminobenzoic  acid  for 
other  essential  processes.  The  double  mutant  will  grow  if  supplied  just  the 
right  amount  of  /(-aminobenzoic  acid  to  satisfy  the  essential  requirements, 
but  not  enough  to  stimulate  the  deleterious  reaction  (Zalokar,  1948). 

Model  Heterocaryons 

It  can  be  seen  that  the  simple  sulfonamide-requiring  mutant  on  the  one 
hand,  and  the  two  double  mutants  on  the  other,  have  different  deficiencies. 
One  produces  methionine  and  ^-aminobenzoic  acid,  but  not  enough  threo- 
nine. The  others  produce  sufficient  threonine,  but  no  methionine,  and  in  one 
case,  no  ^-aminobenzoic  acid.  In  heterocaryons  between  the  simple  and 
double  mutants,  the  two  types  of  nuclei  should  complement  each  other  in 
the  production  of  essential  growth  substances.  If  the  nuclear  ratios  can  be  so 
adjusted  that  the  different  substances  are  produced  in  appropriate  amounts, 
vigorous  growth  should  result.  Heterocaryons  involving  the  simple  sulfona- 
mide-requiring mutant  and  the  double  mutant  sulfonamide-requiring,  amino- 
benzoicless  have  resulted  in  vigorous  growth  (Emerson,  1948)  in  every  test 
so  far  made.  Growth  curves  of  some  of  these  heterocaryons  are  illustrated 
in  Figure  12.6. 

Growth  of  these  heterocaryons  is  usually  not  maintained  at  a  constant 
rate.  Growth  may  stop  completely  after  a  time,  or  it  may  nearly  stop  and 
then  start  again.  This  is  believed  to  be  due  to  fluctuations  in  the  ratio  of  the 
two  kinds  of  nuclei  in  the  advancing  hyphal  tips.  Apparently  there  must  be 
many  times  as  many  double  mutant  nuclei  as  simple  sulfonamide-requiring 
nuclei  to  result  in  a  favorable  combination.  This  is  not  surprising  since  the 


'— I     1)  C 

oB  I 

«     <U  r- 

[1-   X!  00 


sy3i3Piiiim 


212  STERLING  EMERSON 

sulfonamide-requiring  strain  produces  something  in  the  order  of  one  hun- 
dred times  as  much  />-aminobenzoic  acid  as  is  required  for  essential  reactions, 
or  about  fifty  times  as  much  as  is  required  for  the  reaction  which  makes 
threonine  unavailable  for  growth. 

Limited  direct  tests  of  nuclear  frequencies  in  such  heterocaryons  indicate 
that  nuclei  carrying  only  the  sulfonamide-requiring  gene  are  much  less  fre- 
quent than  those  carrying  the  aminobenzoicless  gene  as  well.  In  one  test  of 
about  one  hundred  nuclei,  all  proved  to  be  double  mutants.  In  another  test, 
conidia  from  heterocaryons  were  transferred  to  fresh  growth  tubes  which 
contained  a  concentration  of  sulfanilamide  sufficient  to  inhibit  growth  of  the 
double  mutant  very  strongly  and  still  be  favorable  to  the  growth  of  the  simple 
sulfonamide-requiring  mutant.  Only  one  of  five  such  transfers  grew — again 
suggesting  that  simple  sulfonamide-requiring  nuclei  were  infrequent. 

If  in  order  to  have  rapid  growth  there  must  be  many  double  mutant  nuclei 
and  few  simple  mutants,  it  is  not  surprising  that  vigorous  growth  should 
cease  rather  suddenly.  Ryan,  Beadle,  and  Tatum  (1943)  have  shown  that 
growth  substances  can  be  transported  for  a  distance  of  about  one  centimeter 
in  the  mycelium  of  Neurospora.  One  sulfonamide-requiring  nucleus  at  a  dis- 
tance of  about  a  centimeter  from  the  tip  might  supply  enough  ^-aminoben- 
zoic  acid  for  the  growth  of  that  tip.  But  as  the  tip  grows,  that  nucleus  might 
easily  be  left  behind.  A  deficiency  of  />-aminobenzoic  acid  would  then  de- 
velop in  the  tip,  and  growth  would  be  arrested  unless  a  nucleus  of  the  proper 
constitution  happened  to  migrate  into  the  tip. 

Attempts  to  obtain  rapidly  growing  heterocaryons  involving  the  sulfona- 
mide-requiring mutant  and  the  sulfonamide-requiring,  homocysteineless 
double  mutant  were  unsuccessful.  It  may  be  that  it  is  impossible  to  have  a 
nuclear  ratio  which  will  produce  sufl&cient,  but  not  too  much  methionine, 
and  at  the  same  time  sufficient  threonine  for  the  requirement  of  the  hetero- 
caryon. 

Interpreting  Suppressor  Heterocaryosis  Based  on  Model  Experiments 

The  heterocaryons  between  the  sulfonamide-requiring  mutant  and  its 
double  mutants  with  aminobenzoicless  and  homocysteineless  were  set  up  as 
models  which  should  duplicate  the  behavior  observed  in  the  sulfonamide- 
requiring  strain  when  suppressor  mutations  occurred,  provided  the  interpre- 
tation placed  on  them  was  correct.  For  this  purpose,  the  results  obtained 
were  gratifying.  We  should  like  to  know  just  where  each  of  the  suppressor 
mutations  studied  fits  into  the  biochemical  scheme,  but  at  present  it  can  be 
shown  only  that  they  fit  in  a  general  way. 

Four  suppressors  in  the  first  lot  of  six  (those  illustrated  in  Fig.  12.4), 
which  are  the  only  ones  that  have  been  studied  in  any  detail  at  all,  appar- 
ently represent  mutation  at  four  different  loci,  though  almost  no  direct  tests 


BIOCHEMICAL  MODELS  OF  HETEROSIS  IN  NEUROSPORA  213 

for  allelism  are  available.  The  inference  that  they  are  distinct  genes  is  based 
on  the  data  summarized  in  Table  12.2. 

The  reactions  controlled  by  the  suppressor  genes  have  not  been  identified. 
Suppressor  SA  is  stimulated  in  growth  by  additional  /)-aminobenzoic  acid, 
and  is  inhibited  considerably  by  sulfanilamide  at  concentrations  twenty 
times  less  than  that  required  to  inhibit  wild  type.  It  is  possible  that  a  de- 
ficient amount  of  />-aminobenzoic  acid  is  produced  by  this  mutant,  which 
would  make  it  approximate  the  condition  in  one  of  the  model  heterocaryons. 
Growth  of  suppressor  S-2  is  somewhat  stimulated  by  sulfanilamide  (Fig. 
12.4)  and  by  threonine,  in  this  respect  resembling  the  sulfonamide-requiring 
mutant  which  it  "suppresses."  It  is  even  more  stimulated  by  the  purine, 

TABLE  12.2 

EVIDENCE  SUGGESTING  THAT  SUP- 
PRESSORS 5i,  S2,  Si,  AND  Se  ARE 
DIFFERENT  GENES 


Suppressor 

Second 

Division 

Segregation 

Relation 
to  1633 

Genetically 
Independ- 
ent of 

Si 

25% 

50% 

0% 

60% 

none 

allele  ? 

none 

none 

52 

Si 

s, 

s. 

Si 

adenine,  as  shown  by  the  growth  curves  in  Figure  12.7.  It  was  previously 
known  that  in  the  presence  of  methionine,  adenine  reduces  the  normal  re- 
quirement for  />-aminobenzoic  acid  to  about  one-tenth  its  usual  value.  This 
suggested  that  the  production  of  adenine  also  requires  /)-aminobenzoic  acid. 
The  reaction  controlled  by  this  suppressor  may  thus  be  closely  related  to 
that  controlled  by  the  sulfonamide-requiring  gene.  No  clues  have  turned  up 
to  indicate  how  the  reactions  governed  by  the  remaining  suppressor  muta- 
tions may  be  related  to  these. 

In  the  living  cell  of  Neurospora  the  reactions  which  are  influenced  in  one 
way  or  another  by  the  amount  of  available  /»-aminobenzoic  acid  must  be  fairly 
numerous.  The  production  of  adenine  and  methionine  requires  the  presence 
of  this  vitamin  as  does  the  reaction  in  the  sulfonamide-requiring  mutant 
which  makes  threonine  unavailable. 

Strauss  (1950)  has  studied  a  mutant  strain  (44602)  which  requires  pyri- 
do.xine  unless  grown  at  high  pH  with  ammonia  as  nitrogen  source.  He  found 
that  under  the  latter  conditions  it  is  inhibited  by  methionine,  and  that  this 
inhibition  is  reversed  by  sulfanilamide,  as  if  />-aminobenzoic  acid  were  re- 
quired for  the  inhibition.  Still  another  interrelationship  has  been  found  by 
Shen  (1950)  in  studies  of  a  mutant  strain  (84605)  which  requires  sulfur  in  a 


214 


STERLING  EMERSON 


form  at  least  as  reduced  as  thiosulfate.  At  35°  it  has  no  other  requirement, 
but  at  25°  it  needs  reduced  sulfur,  generally  supplied  as  the  amino  acid 
cysteine,  and  also  tyrosine.  When  methionine  is  supplied  as  the  source  of  sul- 
fur at  25°,  growth  is  strongly  inhibited  by  choline.  Under  these  conditions, 
choline  does  not  inhibit  at  35°,  but  there  is  an  unexpected  stimulation  in 
growth  by  ^-aminobenzoic  acid  at  that  temperature. 

Mutant  strains  have  been  reported  on  two  occasions  which  require  either 
choline  or  />-aminobenzoic  acid — choline  may  be  the  source  of  the  methyl 


to 
ce 


HOURS 

Fig.  12.7 — Growth  curves  of  suppressor  mutant  strain  S-2  on  minimal  medium,  on  threo- 
nine (5  mg/100  ml),  on  methionine,  and  on  purines  (5  mg/100  ml  each  adenine  sulfate  and 

guanine  hydrochloride)  at  35°. 

group  of  methionine.  Strehler  (1950)  has  reported  a  strain  which  requires 
either  methionine  or  /?-aminobenzoic  acid.  There  is  also  a  suggestion  that 
/>-aminobenzoic  acid  may  be  involved  in  the  metabolism  of  lysine.  In  Neuro- 
spora  this  is  suggested  only  because  the  double  mutants  between  the  sul- 
fonamide-requiring  strain  and  two  different  mutants  which  are  unable  to 
synthesize  lysine  do  not  grow  on  any  combination  of  growth  factors  we  have 
tried.  In  bacteria  a  strain  has  been  found  which  requires  either  lysine  or 
/»-aminobenzoic  acid  as  a  growth  factor  (Koft  el  al.,  1950),  strengthening  the 
supposition  of  a  similar  interrelationship  in  Neurospora. 

These  observations  are  referred  to  at  this  time  because  they  indicate  that 
there  are  a  large  number  of  metabolic  reactions  that  are  in  one  way  or  an- 
other related  to  the  availability  of  />-aminobenzoic  acid.  These  reactions 
must  themselves  be  interrelated  in  the  sense  that  an  upset  in  one  of  them 


BIOCHEMICAL  MODELS  OF   HETEROSIS  IN   NEUROSPORA  215 

may  have  a  strong  effect  on  one  or  more  of  the  others,  possibly  through 
changing  the  availability  of  /?-aminobenzoic  acid  or  a  derivative.  The  model 
heterocaryon  experiments  described  earlier  show  that  it  is  possible  for  one 
mutation  to  cause  an  upset  in  one  reaction  and  thus  be  detrimental  to  growth, 
and  for  a  second  mutation  to  restore  conditions  favorable  to  growth  by  actu- 
ally interfering  with  a  different  reaction  which  is  itself  essential  to  growth, 
but  which  is  interrelated  with  the  first  reaction.  In  the  reactions  related  to 
the  metabolism  of  ^-aminobenzoic  acid,  there  is  sufficient  complexity  to  ac- 
count for  the  occurrence  of  a  large  number  of  different  suppressors  of  the  sul- 
fonamide-requiring  character. 

DISCUSSION 

It  has  been  shown  that  increased  vigor  can  result  from  heterocaryosis  in 
which  the  two  kinds  of  nuclei  differ  by  only  one  pair  of  alleles.  This  may  be 
true  only  under  very  special  conditions  such  as  have  been  present  in  the 
examples  discussed.  On  the  other  hand,  it  is  possible  that  the  necessary  con- 
ditions may  be  met  with  rather  frequently  in  Neurospora,  as  suggested  by 
the  following  examples. 

In  mutant  strains  which  have  specific  requirements  for  particular  amino 
acids,  it  is  commonly  found  that  their  growth  is  inhibited  by  the  presence  of 
other  amino  acids  which  do  not  ordinarily  interfere  with  growth.  Some  mu- 
tants which  require  an  outside  source  of  threonine  are  strongly  inhibited  by 
methionine,  (Teas,  Horowitz,  and  Fling,  1948).  Mutants  specifically  requir- 
ing lysine  are  inhibited  by  arginine  (Doermann,  1944),  and  so  on.  In  each  of 
these  instances,  the  inhibition  by  a  particular  amino  acid  is  competitively 
antagonized  by  the  specific  amino  acid  required  by  the  strain  in  question. 
The  growth  of  these  mutants  should  be  favored  by  a  reduction  in  the  amount 
of  the  inhibiting  amino  acid,  as  would  occur  if  some  of  the  nuclei  carried  a 
genetic  block  to  its  synthesis. 

In  extreme  cases,  the  specific  requirement  for  an  amino  acid  may  not  re- 
sult from  a  failure  in  its  synthesis,  but  from  an  oversensitivity  to  the  in- 
hibiting amino  acid.  Thus,  the  sulfonamide-requiring  strain  can  be  said  to 
be  oversensitive  to  homocysteine  in  a  way  that  leads  to  a  requirement  for 
threonine.  One  of  the  lysineless  mutants  (33933)  seems  to  be  oversensitive  to 
arginine  in  much  the  same  way.  Heterocaryons  having  the  lysineless  gene  in 
all  nuclei,  some  of  which  also  carry  a  genetic  block  to  the  synthesis  of  ar- 
ginine (from  strain  36703),  make  considerable  growth  on  minimal  medium, 
whereas  neither  the  lysineless  nor  the  double  mutant  does  (Fig.  12.8). 

Mary  B.  Mitchell  (personal  communication)  recently  observed  that  the 
stock  cultures  of  certain  lysineless  mutants  (4545,  15069,  and  33933)  had 
become  less  sensitive  to  inhibition  by  arginine.  Tests  of  these  showed  that 
they  were  heterocaryons,  some  of  whose  nuclei  were  unchanged.  Some  car- 
ried mutant  genes  which  lowered  the  sensitivity  to  arginine  inhibition  while 


216 


STERLING  EMERSON 


leaving  the  requirement  for  lysine.  These  heterocaryons  were  more  vigorous 
than  the  original  lysineless  strain,  but  no  more  vigorous  than  the  pure  double 
mutant  strains  extracted  from  the  heterocaryons. 

In  studies  on  reverse  mutation  in  a  leucineless  strain  (33757),  Ryan  and 
Lederberg  (1946)  found  that  heterocaryons,  whose  nuclei  differed  only  in 


' 

;| 

•150 

/ 

. 

/                 2', 

• 

*>  i 

/                 f] 

• 

o  / 

/''' 

■ 

TJ   / 

/ 

'1    t 

ex: 

■100 

*»   / 
4.    / 

UJ 

■^/ 

•/    / 

h- 

/t/ 

UJ 

'p' 

-J 
_j 

• 
/ 

/  //J                     • 

S 

/                       / 

/^            /^ 

50 

/ 

/ 

// 

y/ 

/y 

> 

*••/ 

^•^x 

,-' 

.y^  X 

,  -- 

*•  * 

-^l 

^^•-^ 

^   .'y^  ^^%  _ . 

50 

i6o 

HOURS 

Fig.  12.8 — Growth  curves  of  heterocaryons  between  lysineless  (/y,  +)  and  lysineless, 
arginineless  (/y,  arg)  strains  of  Neurospora  at  35°  on  minimal  medium.  Curve  1 :  heterocar- 
yon  in  which  both  nuclear  types  were  of  mating  type  A;  curves  2  to  5:  heterocaryons  made 
up  of  nuclei  of  different  mating  types  (/v,  +,  A  and  /v,  arg,  a) — cf.  Beadle  and  Coonradt 

'   (1944). 


that  some  carried  the  wild  type  allele  and  some  the  mutant  allele  of  the 
leucineless  gene,  almost  invariably  reverted  to  the  homocaryotic  condition. 
By  the  time  growth  had  reached  the  end  of  a  tube  containing  minimal  me- 
dium, nothing  but  wild  type  nuclei  remained.  In  tubes  containing  limiting 
concentrations  of  leucine,  nothing  but  leucineless  nuclei  were  present  after  a 
short  period  of  growth.  This  was  under  conditions  where  the  growth  rate  of 
the  leucineless  strain  is  considerably  less  than  that  of  wild  type.  Under  both 
of  these  conditions,  the  heterocaryon  is  at  a  strong  disadvantage  compared 
to  its  components.  It  is  not  known  whether  or  not  there  is  a  particular  con- 
centration of  leucine  which  would  favor  the  heterocaryon. 


BIOCHEMICAL  MODELS  OF  HETEROSIS  IN  NEUROSPORA  217 

Houlahan  and  Mitchell  (1948)  have  studied  the  interactions  of  mutant 
strains  involved  in  the  metabolism  of  i)yrimidines  and  lysine.  A  pyrimidine- 
less  mutant  (37301)  has  a  specific  requirement  for  pyrimidine.  There  is  a 
suppressor  of  this  mutant  which  enables  it  to  grow  without  added  pyrimidine, 
unless  arginine  is  also  added,  whereupon  the  pyrimidine  requirement  is  re- 
stored. One  lysineless  strain  (33933)  can  utilize  a-amino  adipic  acid  in  place 
of  lysine.  As  a  double  mutant  with  the  pyrimidine  suppressor,  it  can  still 
use  a-amino  adipic  acid,  but  requires  four  times  as  much  as  the  simple  lysine- 
less strain  unless  small  amounts  of  arginine,  or  an  arginine  precursor,  are 
added.  The  double  mutant  combining  this  lysineless  with  the  pyrimidineless 
mutant  is  unable  to  use  a-amino  adipic  acid  unless  a  small  amount  of  lysine 
is  added — arginine  is  ineffective  in  this  instance.  A  second  lysineless  mutant 
(454.5),  which  has  a  specific  requirement  for  lysine  and  which  secretes  pyrimi- 
dines  into  the  medium,  behaves  in  a  predictable  fashion  as  a  double  mutant 
with  pyrimidineless,  or  its  suppressor,  but  not  as  the  triple  mutant  lysineless, 
pyrimidineless,  suppressor  of  pyrimidineless.  Instead  of  requiring  only  lysine 
for  growth,  this  triple  mutant  also  requires  pyrimidines  and  arginine.  This 
example  is  cited  as  another  in  which  metabolic  interactions  may  be  as  com- 
plex as  in  those  discussed  earlier  which  depend  in  one  way  or  another  on 
/>-aminobenzoic  acid. 

Applicability  to  Classical   Heterosis 

Observations  relating  to  one-gene  heterosis  in  higher  plants  are  discussed 
in  other  papers  in  this  series  (Crow,  Hull,  Jones,  and  Whaley).  Studies  of 
Neurospora  heterocaryons  have  shown  that  a  very  similar  phenomenon  oc- 
curs under  certain  special  physiological  conditions.  In  a  particular  genetic 
background,  the  amount  of  an  essential  metabolite  normally  produced  has 
deleterious  consequences  which  are  removed  by  reducing  the  dosage  of  a 
gene  responsible  for  the  production  of  that  metabolite.  This  reduction  was 
brought  about  through  heterocaryosis  in  the  studies  reported,  but  it  should 
also  result  from  heterozygosis  under  similar  physiological  conditions.  There 
is  nothing  in  the  studies  of  heterocaryosis  in  Neurospora  to  suggest  that  one- 
gene  heterosis  is  of  general  occurrence  and  importance,  or  that  other  examples 
should  have  similar  biochemical  backgrounds. 


TH.   DOBZHANSKY 

Columbia  University 


Chapter   13 

Nature  and 
Origin  of  Heterosis 


Exploitation  of  heterosis  in  cultivated  plants  and  animals  is  to  date  by  far 
the  most  important  application  of  the  science  of  genetics  in  agricultural  prac- 
tice. It  is  therefore  unfortunate  that  few  of  the  studies  so  far  made  on 
heterosis  go  beyond  crudely  empirical  observations  and  descriptions  and 
that  little  effort  is  being  made  to  understand  the  underlying  causes  of  the 
phenomena  involved.  Such  an  understanding  is  needed  particularly  because 
the  advances  of  general  genetics  make  it  evident  that  several  quite  distinct, 
and  even  scarcely  related,  phenomena  are  confused  under  the  common  label 
of  heterosis  or  hybrid  vigor. 

In  what  follows,  an  attempt  is  made  to  indicate  briefly  what  seem,  to  the 
writer,  promising  lines  of  approach  to  a  classification  and  study  of  the  various 
kinds  of  heterosis.  The  tentative  nature  of  the  classification  here  suggested 
is  fully  realized.  But  it  is  believed  that  this  classification  may  nevertheless 
serve  a  useful  function  if  it  directs  the  attention  of  the  students  of  heterosis 
to  factors  which  are  only  too  often  overlooked. 

MUTATIONAL  EUHETEROSIS 

Perhaps  the  simplest  kind  of  true  heterosis — euheterosis — is  that  which 
results  from  sheltering  of  deleterious  recessive  mutants  by  their  adaptively 
superior  dominant  alleles  in  populations  of  sexually  reproducing  and  cross- 
fertilizing  organisms. 

Although  only  a  small  fraction  of  the  existing  species  of  organisms  have 
been  investigated  genetically,  it  is  reasonable  to  assume  that  mutational 
changes  arise  from  time  to  time  in  all  species,  albeit  at  different  rates.  Fur- 
thermore, a  great  majority  of  the  mutations  that  arise  are  deleterious,  and 
lower  the  fitness  of  their  carriers  to  survive  or  to  reproduce  in  some  or  in  all 

218 


NATURE  AND  ORIGIN  OF  HETEROSIS  219 

environments.  This  deleterious  character  of  most  mutations  seems  surpris- 
ing, especially  because  in  modern  biology  the  process  of  mutation  is  regarded 
as  the  source  of  the  raw  materials  from  which  evolutionary  changes  are  con- 
structed. 

A  little  consideration  shows,  however,  that  the  ada])tively  negative  char- 
acter of  most  mutations  is  by  no  means  unexpected.  Indeed,  since  every  mu- 
tation has  a  finite  probability  to  occur  in  any  generation,  the  mutants  which 
we  observe  in  our  fields  and  laboratories  must  have  arisen  many  times  in  the 
history  of  the  species.  The  rare  mutants  which  confer  adaptive  advantages 
on  their  possessors  in  the  environments  in  which  the  species  normally  lives 
have  had  the  chance  to  become  established  in  the  species  populations  as 
components  of  the  normal  species  genotype.  In  a  more  or  less  static  environ- 
ment, the  genotypes  of  most  species  are  close  to  the  upper  attainable  level  of 
adaptedness. 

The  above  argument  may  seem  to  prove  too  much.  In  the  absence  of  use- 
ful mutants,  evolution  would  come  to  a  standstill.  The  paradox  is  resolved 
if  we  recall  that  the  environment  is  rarely  static  for  any  considerable  periods 
of  time.  Furthermore,  most  living  species  occur  not  in  a  single  but  in  several 
related  environments.  Genotypes  which  are  adaptively  valuable  in  a  certain 
environment  may  be  ill  adapted  in  other  environments,  and  vice  versa.  It 
should  be  possible  then  to  observe  the  occurrence  of  useful  mutations  if  we 
place  the  experimental  organisms  in  environments  in  which  their  ancestors 
did  not  live. 

Progressive  improvement  of  domesticated  animals  and  plants  in  the  hands 
of  breeders  constitutes  evidence  that  useful  mutations  do  occur.  The  genetic 
variants  which  are  being  made  use  of  by  breeders  have  arisen  ultimately 
through  mutation.  These  mutations  have  been  arising  from  time  to  time,  be- 
fore as  well  as  after  the  domestication.  But  while  they  were  deleterious  in 
the  wild  state,  some  of  them  happened  to  be  suitable  from  the  standpoint  of 
the  breeders.  They  were  useful  in  the  man-made  environment  or  they  were 
useful  to  man.  Favorable  mutations  can  be  observed  also  in  wild  species, 
provided  that  the  latter  are  placed  in  unusual  external  or  genetic  environ- 
ments. This  has  been  demonstrated  in  experiments  of  Spassky  and  the  writer 
on  Drosophila  pseudoobscura.  Several  laboratory  strains  of  this  fly  were  sub- 
jected to  intense  selection  for  fifty  consecutive  generations,  and  improve- 
ments of  the  viability  have  been  observed  in  most  of  them. 

Many,  perhaps  most,  deleterious  mutants  are  nearly  or  completely  reces- 
sive. Others  are  more  or  less  dominant  to  the  "normal,"  or  ancestral,  state. 
The  fate  of  the  dominant  deleterious  mutants  in  jwpulations  of  sexually  re- 
producing and  cross-fertilizing  species  is  different  from  that  of  the  recessives. 
By  definition,  deleterious  mutants  in  wild  species  lower  the  fitness  of  their 
carriers  to  survive  or  to  reproduce,  and  in  cultivated  species  impair  the 
qualities  considered  desirable  by  the  breeders.  Natural  and  artificial  selec- 


220  TH.  DOBZHANSKY 

tion  will  consequently  tend  to  lower  the  frequency,  or  to  eliminate  deleteri- 
ous mutants. 

Selection  against  a  dominant  deleterious  mutant  is,  however,  a  far  more 
efficient  process  than  that  against  a  recessive  mutant.  This  is  because  dele- 
terious recessive  mutant  genes  are  sheltered  from  selection  by  normal  domi- 
nant alleles  in  heterozygotes.  Deleterious  dominants  are  eliminated  by  selec- 
tion within  relatively  few  generations  after  their  origin.  Deleterious  reces- 
sives  accumulate  in  heterozygotes  until  their  frequencies  become  so  high  that 
recessive  homozygotes  are  produced.  Dominant  alleles  are  not  intrinsically 
beneficial,  and  recessives  are  not  necessarily  deleterious.  But  at  any  one  time, 
we  find  in  cross-fertilizing  populations  more  deleterious  recessives  than  dele- 
terious dominants,  because  the  former  are  not  eliminated  by  selection  as 
promptly  as  the  latter. 

Analysis  of  wild  populations  of  several  species  of  Drosophila  has  revealed 
extensive  infestation  of  the  germ  plasm  by  deleterious  recessive  mutant  genes. 
According  to  the  unpublished  data  of  Pavan  and  collaborators,  41  per  cent 
of  the  second  chromosomes  in  Brazilian  populations  of  Drosophila  willistvni 
are  lethal  or  semilethal  when  homozygous.  Among  the  remainder,  57  per 
cent  are  sublethal  when  homozygous.  Furthermore,  31  per  cent  of  the  second 
chromosomes  make  the  homozygotes  completely  sterile  in  at  least  one  sex, 
32  per  cent  retard  the  development,  and  16  per  cent  cause  various  visible 
abnormalities.  Comparable  figures  for  the  third  chromosomes  are  32  per 
cent  of  lethals  and  semilethals,  49  per  cent  subvitals,  28  per  cent  steriles, 
36  per  cent  retarded,  and  16  per  cent  containing  visible  mutants.  Since 
every  fly  has  two  second  and  two  third  chromosomes,  it  is  easily  seen  that  a 
great  majority  of  individuals  in  Brazilian  populations  carry  several  deleteri- 
ous variants  in  heterozygous  condition. 

The  mass  of  deleterious  recessives  carried  in  normally  breeding  natural 
populations  has  no  disastrous  effects  on  the  average  fitness  of  members  of 
such  populations.  This  is  because  the  frequency  of  recessive  homozygotes 
found  in  a  population  at  equilibrium  is  equal  to  the  number  of  the  corre- 
sponding recessive  mutants  that  arise  in  every  generation.  The  loss  of  fitness 
caused  in  a  normally  breeding  population  by  dominant  and  by  recessive  mu- 
tants is  thus  proportional  to  the  frequency  of  the  origin  of  these  mutants  by 
mutation. 

The  situation  changes  completely  if  a  normally  crossbred  population  is 
subjected  to  inbreeding.  For  inbreeding  renders  homozygous  many  reces- 
sives that  would  remain  sheltered  in  heterozygotes  under  normal  crossbreed-  y 
ing.  These  recessives  become  suddenly  exposed  to  natural,  or  to  artificial, 
selection.  The  loss  of  fitness  in  inbred  lines  of  normally  cross-fertilized  species 
is  the  consequence.  Conversely,  the  heterosis  observed  in  the  progeny  of 
intercrossed  inbred  lines  is  the  outcome  of  restoring  the  normal  reproductive 
biology  and  the  normal  population  structure  of  the  species. 


NATURE  AND  ORIGIN   OF   HETEROSIS  221 


BALANCED   EUHETEROSIS 


Balanced  heterosis  is  due  to  the  occurrence  of  a  rather  special  class  of 
mutations  and  gene  combinations,  which  confer  on  heterozygotes  a  higher 
adaptive  value,  or  a  higher  agricultural  usefulness  than  is  found  in  the  cor- 
responding homozygotes. 

The  conditions  most  frequently  found  in  heterozygotes  are  either  domi- 
nance and  recessiveness,  when  the  heterozygote  is  more  or  less  similar  to  one 
of  the  homozygotes,  or  phenotypical  intermediacy  between  the  homozygotes. 
A  heterozygote  may,  however,  be  in  some  respects  phenotypical ly  more  ex- 
treme than  either  homozygote.  Thus,  a  heterozygote  may  be  more  viable, 
more  productive,  or  otherwise  exceed  both  homozygotes  in  some  positive  or 
negative  quality.  This  condition  is  sometimes  spoken  of  as  overdominance 
(Hull). 

Although  overdominance  is,  by  and  large,  an  exceptional  situation,  it  is  of 
particular  interest  to  a  student  of  population  genetics,  and  especially  to  a 
student  of  heterosis.  Suppose  that  a  certain  gene  is  represented  in  a  popula- 
tion by  a  series  of  alleles,  A\  A^,A^  .  .  .  which  are  deleterious  in  homozygous 
condition,  AKi\  A-A^,  A^A^  .  .  .  ,  but  which  show  a  relatively  higher  fitness 
in  heterozygotes  A^A'^,  A'^A^,  A~A^  .  .  .  ,  etc.  Natural  or  artificial  selection 
would  preserve  in  the  population  all  the  variants  A\  A",  A^  .  .  .  ,  regardless 
of  how  poorly  adapted  the  homozygotes  may  be.  In  fact,  one  or  all  homozy- 
gotes may  be  semilethal  or  even  lethal,  and  yet  selection  will  establish  an 
equilibrium  at  which  every  one  of  the  variants  will  be  present  with  a  definite 
frequency.  This  equilibrium  can  easily  be  calculated  if  the  selective  dis- 
advantages of  the  homozygotes,  compared  to  the  heterozygotes,  are  known. 
The  resulting  situation  is  referred  to  as  balanced  polymorphism. 

Balanced  polymorphism  may  be  produced  by  mutations  in  single  genes, 
provided  that  the  heterozygotes  exhibit  overdominance  in  fitness  in  some 
environments.  This  has  been  demonstrated,  among  others,  by  Gustafsson 
and  Nybom.  They  observed  several  mutations  in  barley  that  were  deleterious 
in  homozygotes,  but  produced  heterozygotes  superior  to  the  ancestral  "nor- 
mal" homozygotes.  Ford  and  others  showed  that  certain  color  variants  in 
butterflies,  which  are  inherited  as  though  caused  by  a  single  genetic  change, 
are  maintained  in  natural  populations  by  the  same  mechanism. 

Detailed  data  are  available  on  balanced  polymorphism  in  several  species 
of  Drosophila,  in  which  natural  populations  are  very  often  polymorphic  for 
gene  arrangements  in  some  chromosomes.  These  gene  arrangements  differ  in 
inversions  of  blocks  of  genes.  Thus,  in  certain  populations  of  Drosophila 
pseiidoobscura  from  Southern  California,  at  least  70  per  cent  of  the  wild  indi- 
viduals are  inversion  heterozygotes.  In  populations  of  Drosophila  willisloni 
from  central  Brazil  (Goyaz),  an  average  individual  is  heterozygous  for  as 
many  as  nine  inversions,  and  very  few  individuals  are  homozygous. 


222  TH.  DOBZHANSKY 

Now,  it  has  been  shown  by  observation  both  on  natural  and  on  experi- 
mental populations  of  some  Drosophila  species,  that  the  heterozygotes  for 
the  naturally  occurring  inversions  possess  considerable  adaptive  advantages 
over  the  homozygotes.  For  example,  taking  the  adaptive  value  of  the 
heterozygotes  for  ST  and  CH  inversions  in  Drosophila  pseudoobscura  to  be 
unity,  the  adaptive  values  of  the  ST/ST  and  CH/CH  homozygotes  are 
about  0.8  and  0.4  respectively.  Further,  it  has  been  shown  that  the  heterosis 
in  the  ST/CH  heterozygotes  occurs  only  if  the  constituent  chromosomes  are 
derived  from  the  same  population,  or  from  populations  of  nearby  localities. 
Chromosomes  with  the  same  gene  arrangements,  ST  and  CH,  derived  from 
remote  localities  (such  as  Central  and  Southern  California,  or  Southern 
California  and  Mexico)  exhibit  little  or  no  heterosis. 

This  finding  is  most  compatible  with  the  assumption  that  the  over- 
dominance  in  fitness  observed  in  the  heterozygotes  is  the  property  not  of  a 
single  gene  locus,  or  of  a  chromosome  structure,  but  rather  of  integrated  sys- 
tems of  polygenes.  Such  polygenic  systems  are  coadapted  by  natural  selec- 
tion to  other  polygene  complexes  present  in  the  same  populations.  The  role 
of  the  chromosomal  inversions  in  the  formation  of  the  heterotic  state  of  bal- 
anced polymorphism  is  due  to  the  suppression  of  crossing  over  caused  by 
most  inversions,  at  least  in  Drosophila.  Elimination  of  crossing  over  prevents 
the  breakup  of  the  adaptively  integrated  polygene  complexes  which  are 
carried  in  the  chromosomes  involved. 

It  should  be  noted  that  adaptively  integrated  polygene  complexes  can  be 
maintained  in  crossbreeding  populations  with  the  aid  of  genetic  mechanisms 
other  than  chromosomal  inversions.  Any  factor  which  restricts  or  prevents 
crossing  over  in  chromosomes,  or  parts  of  chromosomes,  can  accomplish  the 
same  biological  function.  Localization  of  chiasmata  may  be  such  a  factor. 
If,  for  example,  chiasmata  are  found  chiefly  or  exclusively  at  some  definite 
points  in  a  chromosome,  the  genes  carried  in  the  sections  which  intervene 
between  these  points  are  inherited  in  blocks.  Such  gene  blocks  may  act 
exactly  as  gene  complexes  bound  together  by  inversions. 

Balanced  heterosis  differs  profoundly  from  mutational  heterosis.  The 
latter  is  due  simply  to  the  sheltering  of  deleterious  recessive  mutants  by 
their  dominant  alleles.  Balanced  heterosis  is  a  result  of  overdominance.  Mu- 
tational heterosis  is  a  protective  device  of  a  sexual  species  with  a  certain 
population  structure  against  the  mutation  pressure.  Balanced  heterosis  is  an 
evolutionary  contrivance  that  permits  maintenance  in  a  population  of  a  mul- 
tiplicity of  genotypes  that  may  be  adaptive  in  dififerent  ecological  niches 
which  the  population  occupies. 

LUXURIANCE 

Mutational  and  balanced  heterosis  resemble  each  other  in  one  important 
respect — both  are  normal  adaptive  states  attained  in  outbred  sexual  species 


NATURE  AND  ORIGIN  OF  HETEROSIS  223 

as  a  result  of  an  evolutionary  history  controlled  by  natural  or  by  artificial 
selection.  The  normal  heterotic  state  can  be  disru])ted  by  sudden  inbreeding, 
which  is  evidently  a  disturbance  of  the  reproductive  biology  to  which  the 
species  is  adjusted.  The  heterotic  state  can  also  be  restored  by  intercrossing 
the  inbred  lines.  This  is  true  heterosis,  or  euheterosis.  Euheterosis  is  a  form 
of  evolutionary  adaptation  characteristic  of  sexually  reproducing  and  cross- 
fertilizing  species. 

Numerous  instances  are  known,  however,  when  hybrids  between  si)ecies, 
neither  of  which  can  be  regarded  as  inbred,  are  larger,  faster  growing,  or 
otherwise  exceeding  the  parental  forms  in  some  quality.  Similar  luxuriance  is 
observed  in  some  hybrids  between  normally  self-fertilizing  species,  races,  or 
strains.  This  kind  of  luxuriance  of  hybrids  cannot  be  ascribed  to  sheltering  of 
deleterious  recessive  mutants,  because  the  latter  are  sheltered  in  the  parental 
populations.  It  is  also  unlikely  to  arise  from  overdominance  since,  at  least  in 
wild  species,  natural  selection  would  be  expected  to  have  induced  such  bal- 
anced heterosis  in  the  parental  species  or  strains. 

Luxuriance  is,  from  the  evolutionary  standpoint,  an  accidental  condition 
brought  about  by  complementary  action  of  genes  found  in  the  parental  form 
crossed.  Two  sets  of  facts  are  important  in  this  connection.  First,  in  cases  of 
luxuriance  there  is  usually  no  indication  whatever  that  the  luxuriant  hybrids 
would  prove  adaptively  superior  in  competition  with  the  parental  forms  in 
the  natural  habitats  of  the  latter.  Second,  luxuriance  appears  to  be  more 
frequently  encountered  in  domesticated  than  in  wild  species. 

It  stands  to  reason  that  increase  in  body  size,  or  in  growth  rate,  is  by  no 
means  always  an  adaptively  superior  change.  To  equate  size  with  vigor,  fit- 
ness, or  adaptive  value  would  be  a  height  of  anthropomorphic  naivete.  The 
rate  of  growth  and  the  size  attained  by  an  organism  in  its  normal  environ- 
ments are  evidently  controlled  by  natural  selection.  Excessive  as  well  as  de- 
ficient sizes  are  adaptively  about  equally  disadvantageous.  The  checks  upon 
excessively  rapid  growth  and  excessive  size  are,  however,  very  often  relaxed 
under  domestication.  In  man-controlled  environments  those  qualities  often 
become  desirable  from  the  standpoint  of  the  breeder  if  not  from  that  of  the 
organism.  Luxuriance  is,  really,  pseudoheterosis. 


DONALD  F.  JONES 

Connecf/'cuf  Agricultural  Exper'imenf  Sfaiion 


Chapter   14 

Plasmagenes  and 
Chromogenes  in  Heterosis 


The  word  heterosis  is  essentially  a  contraction  of  the  phrase  stimulus  of 
heterozygosis.  It  was  first  used  by  G.  H.  Shull  (1914).  The  concept  of  a 
stimulation  resulting  from  the  genetic  union  of  unlike  elements  was  de- 
veloped by  East  (1909).  Previous  to  the  Mendelian  conception  of  units  of 
heredity,  it  was  generally  considered  by  plant  and  animal  breeders  that  the 
invigorating  effect  of  crossing  unlike  varieties  of  plants  and  breeds  of  live- 
stock was  due  to  the  correction  of  imperfections  that  existed  in  both  parental 
types.  This  idea  is  clearly  stated  by  Samuel  Johnson  in  the  second  edition  of 
his  book  How  Crops  Grow  (1891). 

The  early  recordings  of  instances  of  hybrid  vigor  and  the  various  means 
of  accounting  for  this  phenomenon  have  been  stated  and  restated  so  many 
times  that  there  is  no  need  or  useful  purpose  in  repeating  them  here.  Excel- 
lent reviews  of  the  literature  are  readily  available  (see  especially  East  and 
Hayes,  1912;  Jones,  1918;  East  and  Jones,  1919;  East,  1936;  and  Whaley, 
1944). 

THE  EXPRESSION  OF  HETEROSIS 

At  the  present  time,  the  term  heterosis  designates  the  increased  growth  or 
other  augmented  action  resulting  from  crossing,  however  it  is  produced.  As 
generally  used,  it  is  essentially  synonymous  with  hybrid  vigor.  Heterosis  has 
two  general  modes  of  expression.  In  one,  there  is  an  increase  in  size  or  num- 
ber of  parts.  This  is  usually  the  result  of  a  greater  number  of  cells  and  a  faster 
rate  of  cell  division  and  cell  activities.  This  results  in  an  improvement  in  gen- 
eral well-being  of  the  organism  similar  to  the  result  of  being  placed  in  a  more 
favorable  environment.  Such  luxuriance  may  be  accompanied  by  partial  or 
complete  sterility  in  diverse  crosses. 

A  somewhat  different  manifestation  of  heterosis  is  an  increase  in  bio- 

224 


PLASMAGENES  AND  CHROMOGENES  IN   HETEROSIS  225 

logical  efficiency,  such  as  reproductive  rate  and  survival  ability.  This  may 
even  be  shown  with  a  reduction  in  productiveness  as  measured  by  economic 
characters.  Some  confusion  has  arisen  by  not  distinguishing  clearly  between 
these  two  different  manifestations  of  heterosis. 

In  addition  to  these  two  general  types  of  heterotic  effects,  there  may  also 
be  a  reduction  in  both  growth  and  survival  ability;  in  other  words,  hybrid 
weakness  or  a  reversed  or  negative  heterosis.  This  effect  is  much  less  com- 
mon and  is  seldom  found  in  cultivated  plants  and  domesticated  animals. 

TYPES  OF  GENE  ACTION 

An  understanding  of  the  mode  of  action  of  heterosis  has  now  resolved  into 
a  study  of  the  nature  of  gene  action.  The  genes  usually  used  to  illustrate 
Mendelism  are  the  loss  variations  that  have  a  major  effect  such  as  the  inabil- 
ity to  produce  some  essential  substance.  This  results  in  a  block  in  the  normal 
chemical  processes,  finally  resulting  in  an  individual  of  greatly  altered  ap- 
pearance, size,  or  ability  to  survive.  The  effect  ranges  in  intensity  from  a  com- 
pletely lethal  condition  at  some  stage  of  development,  up  to  individuals  that 
differ  only  slightly  in  appearance  from  normal  with  no  appreciable  reduction 
in  growth  or  survival  ability.  Such  genes  are  illustrated  by  the  long  lists  of 
Mendelizing  characters  now  tabulated  for  maize,  Drosophila,  mice,  and  many 
other  animals,  plants,  and  lower  organisms. 

DOMINANT  AND  RECESSIVE  GENES 

In  these  cases,  the  normal  allele  is  usually  designated  by  a  capital  letter, 
with  the  mutant,  deficient  allele  denoted  by  the  corresponding  lower  case 
letter.  In  comparison  with  the  normal  allele,  the  recessive  mutants  are  de- 
ficient in  some  respect.  In  their  inability  to  produce  certain  specific  sub- 
stances, as  shown  in  the  haploid  Neurospora  by  Beadle  and  his  co-workers, 
they  are  referred  to  as  .4 -less,  5-less,  C-less,  etc.  In  diploid  organisms  A  is 
usually  completely  dominant  over  a;  that  is,  one  A  allele  functions  as  well 
or  nearly  as  well  as  two. 

There  is  no  question  that  the  accumulation  in  a  hybrid  of  the  normal 
alleles  of  this  type  results  in  heterosis.  In  the  simplest  example  of  a  cross  of 
^-less  by  B-less  (aaBB  XAAbb)  the  hybrid  offspring  are  all  AaBb,  and 
essentially  normal  for  whatever  effect  A  and  B  have.  But  since  the  mutant 
recessive  alleles  of  this  type  are  so  drastic  in  their  effect,  most  of  these 
deficiencies  are  removed  by  natural  selection  in  all  species  whether  self- 
fertilized  or  cross-fertilized.  Therefore  they  have  little  part  in  the  heterosis 
that  is  shown  by  these  organisms  when  crossed.  Furthermore,  genes  of  this 
type  are  eliminated  when  naturally  cross-fertilized  species,  such  as  maize, 
are  artificially  self-pollinated.  Yet  such  inbred  strains  show  the  largest 
amounts  of  heterosis. 

There  is  evidence,  as  will  be  shown  later,  that  there  are  many  genes  of  this 


226  DONALD  F.  JONES 

type  having  small  effects  that  are  not  eliminated  by  natural  or  artificial 
selection  either  in  the  wild  or  under  domestication,  and  that  these  deficiencies 
or  degenerative  mutants  do  have  a  large  part  in  bringing  about  reduced 
growth.  Before  presenting  this  evidence,  there  are  other  types  of  gene  action 
that  should  be  considered. 

CHROMOSOMAL  DELETIONS 
In  addition  to  the  recessive  mutant  alleles  that  are  deficient  as  compared 
to  their  normal  alleles,  there  are  also  chromosomal  deletions  which  result  in 
the  complete  elimination  of  the  normal  locus.  Large  deletions  are  usually 
lethal  and  are  quickly  eliminated.  Small  deletions  that  cannot  be  detected 
cytologically  are  haplo-viable,  and  may  persist  indefinitely  if  they  are  closely 
linked  with  essential  loci.  Changes  of  this  type  have  been  demonstrated  by 
McClintock  (1931)  and  by  Stadler  (1933).  They  cannot  be  readily  distin- 
guished from  recessive  mutants  of  the  .4 -less  type.  In  fact  there  may  be  no 
difference.  In  practically  all  cases  they  show  varying  amounts  of  germ  cell 
abortion,  and  do  not  mutate  back  to  normal.  Deletions  of  this  type  are 
designated  Ao. 

DOMINANT   UNFAVORABLE  GENES 

In  many  cases  of  deletion  the  heterozygote,  or  the  hemizygote,  is  visibly 
and  unfavorably  altered  from  normal,  in  which  event  the  genes  involved  are 
listed  as  dominant,  and  if  partially  viable  they  can  bring  about  negative 
heterosis  or  hybrid  weakness.  It  is  not  known  whether  all  dominant  unfavor- 
able genes  are  deletions  of  this  type,  but  as  far  as  their  effect  on  heterosis  is 
concerned  it  makes  little  difference  whether  or  not  they  are.  An  illustration 
of  this  type  of  gene  action  may  be  seen  in  a  cross  of  Ragged  and  Knotted 
maize  plants.  Both  of  these  genes  result  in  a  marked  reduction  in  growth  in 
the  heterozygous  condition.  They  are  not  completely  lethal  in  the  homozy- 
gous dominant  condition,  but  seldom  produce  seed  or  pollen.  When  both 
dominant  genes  are  present  together  in  the  heterozygous  condition,  there  is 
a  marked  reduction  in  size,  rate  of  growth,  and  reproductive  ability  as  com- 
pared with  either  parental  type. 

Tunicate,  teopod,  and  corn  grass  are  also  dominant  genes  that  reduce 
grain  yields  in  both  the  homozygous  and  heterozygous  condition.  They  are 
probably  reversions  to  a  primitive  condition  which  in  suitable  genetic  com- 
binations maybe  favorable  to  survival  in  the  wild.  Dunn  and  Caspari  (1945) 
describe  many  structural  abnormalities  in  mice  that  seem  to  be  due  to  dele- 
tions having  a  dominant  effect  in  the  hemizygote.  Some  of  these  counteract 
each  other  and  tend  to  restore  a  more  normal  condition,  while  others  accumu- 
late unfavorable  effects.  A  similar  situation  has  been  reported  in  Drosophila 

by  Stern  (1948).  . 

In  addition  to  recessive  deletions  with  a  dominant  effect  m  the  hetero- 
zygote, there  are  also  dominant  inhibitors  that  have  no  indication  of  being 


PLASMAGENES  AND  CHROMOGENES  IN   HETEROSIS  227 

deletions,  but  do  prevent  other  genes  from  liaving  their  usual  exi)ression. 
Most  of  these  inhibitors  control  color  characters  and  are  usually  not  involved 
in  heterosis.  If  they  were,  there  would  be  more  negative  heterosis  than  actual- 
ly is  found. 

GENES  WITHOUT  DOMINANCE 

Unlike  the  visible  Mendelizing  genes  with  their  clear-cut  dominance  and 
unfavorable  action  of  one  or  the  other  allele,  there  are  many  genes  that  dif- 
ferentiate size  or  number  of  parts,  time  of  flowering  and  maturing.  These  are 
the  genes  usually  involved  in  normal  variation.  They  are  the  ones  the  plant 
and  animal  breeder  are  mainly  concerned  with  and  could  expect  to  have  a 
major  effect  on  heterosis.  Since  neither  member  of  an  allelic  pair  can  be  con- 
sidered abnormal  or  deficient,  both  are  designated  with  a  capital  letter  with 
some  prefix  to  differentiate  them,  as  for  example  A  and  A\ 

Genes  of  this  type  usually  have  simple  additive  effects  such  as  the  F  endo- 
sperm color  gene  in  maize,  in  which  each  allele  adds  a  definite  increment  in 
total  carotene  content.  Such  additive  genes  without  dominance  are  used  to 
interpret  the  inheritance  of  quantitative  characters  which  have  been  shown 
to  segregate  and  recombine  in  a  Mendelian  manner. 

No  clear  distinction  can  be  made  between  the  A  a  and  .4^1'  types  of  genes 
and  this  has  led  to  much  confusion.  The  first  class  shows  complete  or  nearly 
complete  dominance.  The  second  shows  no  dominance  or  very  little  domi- 
nance, but  one  type  integrates  into  the  other.  The  principal  question  at  issue 
is  whether  either  type  shows  over-dominance,  or  in  other  words,  an  interac- 
tion between  alleles  such  that  Aa  >  A  A  or  aa  or  AA^  >  A  A  or  .4 '.4'.  Before 
considering  the  evidence  for  or  against  over-dominance,  two  remaining  types 
of  genes  should  be  considered. 

CHROMOSOMAL  REARRANGEMENTS 

By  chromosomal  rearrangements  such  as  inversions  and  translocations, 
genes  without  alteration  are  placed  in  different  spatial  relations  with  other 
genes.  In  their  altered  position  they  have  different  effects.  Dobzhansky  and 
his  associates  have  studied  many  geographical  races  of  Drosophila  that  differ 
by  chromosomal  rearrangements.  Crosses  between  these  chromosomal  types 
from  the  same  region  exhibit  heterosis,  whereas  the  same  chromosomal  type 
from  different  regions  do  not  show  such  a  high  degree  of  heterosis.  This 
seems  not  to  be  a  position  effect,  but  is  the  result  of  an  accumulation  of  gene 
differences  that  are  protected  from  random  distribution  by  the  prevention  of 
crossing  over  in  hybrids  of  different  chromosomal  types. 

COMPOUND  GENES  AND  GENES  WITH  MULTIPLE  EFFECTS 
In  many  organisms,  loci  are  known  which  have  different  effects  on  differ- 
ent parts  of  the  organism.  In  maize  the  .1 ,  P,  and  R  genes  have  been  studied 
in  considerable  detail  by  Stadler  and  his  co-workers.  These  loci  each  have  a 


228  DONALD  F.  JONES 

series  of  alleles  that  produce  characteristic  color  patterns  and  intensities  of 
colors  in  different  parts  of  the  plant  such  as  culm,  leaf  sheath,  leaf  blade, 
glumes,  anthers,  silks,  cob  and  pericarp,  and  endosperm.  They  may  be  con- 
sidered either  as  genes  located  so  closely  together  that  they  never  show- 
crossing  over,  or  compound  genes  with  multiple  effects.  Without  going  into 
the  evidence  for  or  against  these  two  hypotheses,  it  is  obvious  that  compound 
genes  can  have  an  important  part  in  heterosis  if  they  control  growth  proc- 
esses. More  information  is  needed  on  the  specific  effect  of  compound  genes. 

In  Godetia  a  series  of  multiple  alleles  has  been  described  by  Hiorth  (1940) 
that  is  often  cited  as  an  illustration  of  an  interaction  between  alleles  produc- 
ing an  effect  analogous  to  heterosis.  Actually  these  are  color  determiners  that 
control  pigment  production  in  different  parts  of  the  flower  quite  similar  to  the 
A,  P,  and  R  loci  in  maize.  Each  allele  has  a  different  manifestation,  and  all 
tend  to  accumulate  color  in  the  heterozygotes. 

The  familiar  notation  of  a  chromosome  as  a  linear  arrangement  of  loci, 
each  of  which  is  the  site  of  a  single  gene  with  one  effect  function,  is  probably 
an  oversimplification  of  the  actual  condition.  It  is  difficult  to  see  how  an 
organism  could  have  originated  in  this  way.  It  is  more  likely  that  a  chromo- 
some is  an  association  of  primitive  organisms  of  varying  types  and  functions. 
These  primitive  organisms  found  it  to  be  an  advantage  in  the  evolutionary 
process  to  become  associated  in  some  such  process  as  the  colonial  organisms 
now  exhibit.  This  association  has  undergone  very  great  modification  and 
ramifications,  but  the  compound  genes  may  be  vestigial  structures  of  such 
an  association,  differing  greatly  in  size,  arrangement,  and  function.  Many  of 
them  still  retain  some  independence,  and  when  removed  from  their  normal 
position  in  the  chromosome  could  function  as  plasmagene  or  viroid  bodies. 

These  compound  genes  may  undergo  mutation  and  possibly  recombina- 
tion or  reorganization  within  themselves,  but  crossing  over  takes  place  for 
the  most  part  only  between  these  compound  structures.  Compound  genes 
also  arise  by  unequal  crossing  over  and  duplication  of  loci  are  shown  by  the 
Bar  eye  gene  in  Drosophila  and  others  of  similar  type. 

In  addition  to  compound  or  multiple  genes,  there  are  single  genes  with 
multiple  effects.  Many  of  these  are  important  in  growth  processes  and  are 
illustrated  by  chlorophyll  production  in  maize  studied  by  H.  L.  Everett 
(1949).  One  major  gene  is  essential  for  the  production  of  carotene.  In  the 
recessive  condition  the  seeds  are  pale  yellow  in  color,  in  a  normal,  dark  yel- 
low seeded  variety.  The  young  seedlings  grown  from  these  pale  yellow  seeds 
are  devoid  of  chlorophyll.  The  recessive  allele  is  therefore  lethal.  By  using  the 
pale  yellow  endosperm  as  a  convenient  marker  and  crossing  with  a  number  of 
standard  field  corn  inbreds,  it  has  been  found  that  these  inbreds  differ  widely 
in  their  normal  chorophyll  mechanism.  Many  of  them  have  genes  that  can 
restore  normal  chorophyll  production  without  restoring  the  production  of 
carotene  in  the  seed.  Other  genes  restore  chlorophyll  production  only  partial- 


PLASMAGENES  AND  CHROMOGENES   IN   HETEROSIS 


229 


]y  (see  Table  14.1).  Hybrid  combinations  that  bring  these  genes  together  are 
appreciably  more  efficient  in  chl()ro])hyll  production  than  combinations  that 
lack  some  of  them.  However  one  of  these  dominant  alleles  has  a  suppressing 
eflfect  on  chlorophyll  development.  The  combination  of  all  of  these  chloro- 
phyll genes  so  far  studied  is  not  the  most  productive.  There  are  many  genes 
of  this  type  that  block  chemical  syntheses,  that  are  not  lethal  in  the  usual 
genetic  assembly,  but  which  combine  to  give  a  cumulative  efficiency  in  most 
cases. 

Lethal  genes  which  show  complete  dominance  of  the  normal  allele  would 
have  no  effect  on  heterosis  other  than  to  reduce  the  number  of  offspring.  Such 

TABLE  14.1 

GENES  CONTROLLING  CHLOROPH\TL  PRO- 
DUCTION IN  MAIZE* 


Ch 

Ch 

Ch 

Seed  Color 

Chlor.  Grade 

Viability 







Pale 

Albino 

Lethal 

— 

+ 

_ 

Pale 

Virescent 

Lethal 

— 

+ 

Pale 

Light  green 

Normal 

— 

+ 

+ 

Pale 

Light  green 

Normal 

+ 

— 

— 

Yellow 

Light  green 

Normal 

+ 

+ 

— 

Yellow 

Med.  green 

Normal 

+ 

+ 

+ 

Yellow 

Dark  green 

Normal 

+ 

+ 

Yellow 

? 

Normal 

*  Data  from  H.  L.  Everett. 


genes  would  be  just  as  effective  in  the  homozygous  as  the  heterozygous  con- 
dition. Genes  that  have  any  part  in  the  type  of  heterosis  that  is  manifested 
in  increased  growth  must  be  viable  and  have  some  degree  of  dominance.  In 
other  words,  Aa  must  be  greater  than  |  AA.  Aa  may  even  be  greater  in 
its  effect  than  A  A  or  aa  in  which  case  theoretically  there  is  over-dominance, 
but  very  little  specific  evidence  is  available  to  show  that  such  a  situation 
actually  exists. 

I  can  see  no  way  in  which  it  is  possible  to  separate  over-dominance  from 
a  stimulus  of  heterozygosis.  They  seem  to  be  different  ways  of  saying  the 
same  thing.  The  essential  point  at  issue  at  the  present  time  is  whether  or  not 
over-dominance  actually  occurs,  and  if  so,  how  important  this  is  in  the 
total  amount  of  heterosis  in  addition  to  the  known  accumulation  of  favorable 
dominant  effects. 

INTERACTION   BETWEEN  ALLELES 

Evidence  has  been  presented  from  many  sources  bearing  on  the  problem 
of  over-dominance  and  interaction  between  different  alleles.  Much  of  the 
argument  is  based  on  mathematical  treatment  of  data  that  require  many  as- 
sumptions. What  is  needed  is  more  specific  evidence  where  the  effect  of 


230  DONALD  F.  JONES 

multiple  genes  can  be  ruled  out.  Very  few  specific  examples  of  single  gene 
action  are  available. 

In  one  case  studied  by  the  writer  there  is  clear  evidence  for  an  interaction 
between  alleles  (Jones,  1921).  A  mutation  in  a  variety  of  normally  self-ferti- 
lized tobacco  changed  a  determinate  plant  into  an  indeterminate,  non- 
flowering  variation.  It  was  a  change  in  the  normal  response  to  the  summer 
day  length  period.  The  mutant  plants  failed  to  flower  in  the  normal  growing 
season  and  continued  in  a  vegetative  condition.  Reciprocal  crosses  between 
the  mutant  and  normal  types  both  grew  at  the  same  rate  as  the  normal 
plants  showing  complete  dominance  of  the  normal  growth  rate.  The  hetero- 
zygous plants  continued  their  vegetative  growth  longer  and  produced  taller 
plants  with  more  leaves  and  flowers  than  the  normal  homozygous  plants. 
This  result  I  consider  not  to  be  heterosis,  since  there  was  no  increase  in 
growth  rate.  It  is  merely  an  interaction  between  alleles  to  produce  a  result 
that  is  different  from  either  parent.  There  are  undoubtedly  many  allelic 
interactions  of  this  type.  Whether  or  not  they  can  be  considered  to  contribute 
to  heterosis  is  largely  a  matter  of  opinion. 

Other  cases  in  corn  where  heterosis  resulted  from  degenerative  changes 
(Jones,  1945)  were  at  first  assumed  to  be  single  allelic  differences,  since  they 
originated  as  mutations  in  inbred  and  highly  homozygous  families.  The  de- 
generate alterations  were  expressed  as  narrow  leaves,  dwarf  plants,  crooked 
stalks,  reduced  chlorophyll,  and  late  flowering.  All  of  these  mutant  variations 
gave  larger  amount  of  growth  in  a  shorter  period  of  time  and  clearly  showed 
heterosis. 

The  further  study  of  this  material  has  not  been  completed,  but  the  results 
to  date  indicate  that  the  differences  involved  are  not  single  genes.  Both  the 
extracted  homozygous  recessives  and  the  extracted  homozygous  dominants 
from  these  crosses  are  larger  than  the  corresponding  plants  that  originally 
went  into  the  crosses. 

This  indicates  quite  clearly  that  the  visible  changes  were  accompanied  or 
preceded  by  other  changes  with  no  noticeable  effects,  but  which  are  expressed 
in  growth  rates.  A  more  complete  summary  of  these  results  will  have  to 
wait  until  all  of  the  evidence  is  at  hand.  It  is  a  simple  matter  to  extract 
the  homozygous  recessives  from  these  crosses,  but  it  is  difficult  to  extract 
the  homozygous  dominants.  Many  of  the  self-fertilized  plants  proved  to  be 
heterozygous. 

GENES  CONTROLLING  GROWTH 

Additional  evidence  that  there  are  a  large  number  of  genes  having  small 
effects  on  growth  without  visible  morphological  changes  is  becoming  clearly 
apparent  from  a  backcrossing  experiment  now  in  progress.  Several  long 
inbred  lines  of  corn,  one  of  which  is  now  in  the  forty-first  generation  of  con- 
tinuous self-fertilization,  were  outcrossed  to  unrelated  inbred  lines  having 
dominant  gene  markers  which  could  be  easily  selected.  The  markers — red 


PLASMAGENES  AND  CHROMOGENES  IN  HETEROSIS  231 

cob,  yellow  endosperm,  and  non-glossy  seedlings—  were  chosen  because  Ihey 
had  little  or  no  effect  on  growth  of  the  plant. 

The  first  generation  outcrosses  showed  the  usual  large  increases  in  size  of 
plant,  time  of  flowering,  and  yield  of  grain  that  is  expected  in  crosses  of  un- 
related inbred  strains  of  corn.  The  hybrid  plants  were  backcrossed  as  seed 
parents  with  pollen  from  the  inbred  with  the  recessive  gene  marker.  In  every 
generation,  plants  with  the  dominant  gene  marker  were  selected  for  back- 
crossing.  These  plants  have  now  been  backcrossed  six  successive  times.  Many 
progenies  have  been  grown.  They  are  all  heterozygous  for  the  gene  marker 
plus  whatever  neighboring  regions  on  the  same  original  chromosome  from 
the  non-recurrent  parent  that  have  not  been  lost  by  crossing  over. 

The  plan  is  to  continue  the  backcrossing  until  no  measurable  differences 
remain  between  the  backcrossed  plants  and  the  recurrent  parent,  or  be- 
tween the  two  classes  of  backcrossed  individuals  in  the  same  backcrossed 
progeny,  those  with  the  dominant  marker  and  those  with  the  recessive 
marker.  When  the  point  is  reached  where  no  differences  can  be  detected,  the 
plan  is  to  compare  successive  earlier  generations  from  remnant  seed  to  pick 
up  whatever  single  gene  differences  there  might  be  that  could  be  measured 
and  detected  by  their  segregation. 

So  far  both  classes  of  backcrossed  plants  in  nearly  all  progenies  are  taller 
and  flower  earlier,  showing  that  they  have  not  been  completely  converged  to 
the  parental  type  (see  Table  14.2).  The  differences  are  small  and  not  statisti- 
cally significant  in  the  tests  so  far  made,  but  are  nearly  all  in  the  direction  of  a 
heterotic  effect.  As  yet  there  are  not  sufiicient  data  to  base  final  conclusions. 
It  is  hoped  that  the  comparison  of  the  two  classes  of  backcrossed  progeny 
with  the  original  recessive  parent  will  permit  a  distinction  between  the  favor- 
able action  of  dominant  genes  and  an  interaction  between  heterozygous 
alleles.  Also  that  it  may  be  possible  to  show  whether  or  not  there  is  any 
residual  cytoplasmic  effect,  since  some  of  the  outcrossed  plants  have  the  same 
cytoplasm  as  the  dominant  gene  marker  and  some  do  not. 

Important  facts  do  stand  out  clearly  from  this  experiment.  Since  heterosis 
still  remains  after  these  many  generations  of  backcrossing,  it  shows  clearly 
that  these  three  chromosome  regions  selected  as  samples  have  an  appreciable 
effect  on  growth.  Since  the  gene  markers  themselves  have  no  effect  on 
growth,  as  far  as  this  can  be  determined  in  other  material,  these  three  regions 
are  random  selections  for  growth  effects.  This  indicates  quite  clearly  that 
there  are  genes  in  all  parts  of  the  chromosomes  that  contribute  to  normal 
growth  and  development.  While  the  evidence  so  far  available  does  not  per- 
mit a  clear  separation  between  the  effects  of  an  accumulation  of  favorable 
genes  as  contrasted  to  an  interaction  between  alleles,  or  between  genes  and 
cytoplasm,  the  results  show  that  there  are  many  loci  involved  in  the  heterotic 
effect  in  addition  to  the  dominant  gene  markers. 

This  follows  from  the  evidence  at  hand.  If  the  heterosis  now  remaining 


232  DONALD  F.  JONES 

were  due  solely  to  the  interaction  between  the  dominant  and  recessive  mark- 
ers, there  would  have  been  a  rapid  approach  to  the  level  of  vigor  now  re- 
maining. If  it  were  due  to  a  larger  number  of  genes  distributed  rather  evenly 
along  the  chromosome,  the  reduction  in  heterosis  would  be  gradual,  as  it 
has  proved  to  be.  Small  amounts  of  heterosis  may  persist  for  a  long  time  un- 
til all  of  the  genes  contributing  to  it  are  removed  by  crossing  over. 

A  recent  experiment  by  Stringfield  (1950)  shows  a  difference  in  produc- 
tiveness between  an  F2  selfed  generation  and  a  backcross  having  the  same 
parentage.  The  amount  of  heterozygosis  as  measured  by  the  number  of 
allelic  pairs  is  the  same  in  both  lots.  In  the  backcross  there  are  more  indi- 
viduals in  the  intermediate  classes  with  respect  to  the  number  of  dominant 

TABLE  14.2 

INCREASE  IN  HEIGHT  OF  PLANT  IN  SUCCES- 
SIVE BACKCROSSED  GENERATIONS  HET- 
EROZYGOUS FOR  A  DOMINANT  GENE 
MARKER 


Number  of 

Per  Cent  Increase  in  Height 

Generations 
Backcrossed— > 

4 

5 

6 

20yX243Y 

20yXP8  Y 

20pX243P 

243glX20Gl 

6.7 
1.9 
6.6 

2.2 
2.3 
3.0 

1.5 

1.2 

1.1 

-1.3 

genes.  This  indicates  a  complementary  action  of  favorable  dominant  genes. 
Gowen  et  al.  (1946)  compared  the  differences  in  egg  yield  in  Drosophila 
between  random  matings,  47  generations  of  sib  mating,  and  homozygous 
matings  by  outcrossing  with  marker  genes.  The  differences  are  significant, 
and  indicate  a  large  number  of  genes  having  dominant  effects  on  the  repro- 
ductive rate. 

INTERACTION  OF  GENES  AND  CYTOPLASM 

The  suggestion  has  been  made  many  times  that  heterosis  may  result  from 
an  interaction  between  genes  and  cytoplasm.  Within  the  species,  differences 
in  reciprocal  crosses  are  rare.  In  commercial  corn  hybrids,  reciprocal  differ- 
ences are  so  small  that  they  can  usually  be  ignored.  Evidence  is  accumulating 
that  there  are  transmissible  differences  associated  with  the  cytoplasm,  and 
that  these  must  be  considered  in  a  study  of  heterosis.  Small  maternal  effects 
are  difficult  to  distinguish  from  nutritional  and  other  influences  determined 
by  the  genotype  of  the  mother  and  carried  over  to  the  next  generation. 

The  cross  of  the  two  different  flowering  types  of  tobacco  previously  cited 
shows  a  maternal  effect.  The  first  generation  cross  of  the  indeterminate  or 


PLASMAGENES  AND  CHROMOGENES  IN   HETEROSIS  233 

non-flowering  type  as  seed  parent  grows  taller  than  the  reciprocal  com- 
bination, and  flowers  later.  These  difi"erences  are  statistically  significant. 

Reciprocal  crosses  between  inbred  California  Rice  pop,  having  the  small- 
est seeds  known  in  corn,  with  inbred  Indiana  Wf9  having  large  embryos 
and  endospersms,  show  differences  in  early  seedling  growth  and  in  tillering. 
Inbred  Wf9  produces  no  tillers.  California  Rice,  also  inbred,  produces  an 
average  of  4.1  tillers  per  plant.  The  first  generation  cross  of  Rice  popXWfQ 
averages  1.0  tillers,  while  the  reciprocal  combination  under  the  same  condi- 
tions produced  2.2  tillers  per  stalk.  In  this  case  the  non-tillering  variety, 
when  used  as  the  seed  parent,  produces  more  than  twice  as  many  tillers. 
This  seems  to  be  a  carry-over  effect  of  the  large  seed.  Tillering  is  largely  de- 
termined by  early  seedling  vigor.  Anything  that  induces  rapid  development 
in  the  early  stages  of  growth  tends  to  promote  tillering. 

PLASMAGENES  AND  CHROMOGENES 

In  addition  to  these  transitory  effects  there  are  many  cases  of  cytoplasmic 
inheritance.  Caspari  (1948)  has  reviewed  the  evidence  from  fungi,  mosses, 
the  higher  plants,  and  from  Paramecium,  insects,  and  mammals  to  show 
that  many  differences  do  occur  in  reciprocal  crosses  and  that  they  persist  into 
later  backcrossed  generations.  Reciprocal  differences  in  the  amount  of 
heterosis  have  been  demonstrated  in  Epilobium  (Michaelis,  1939)  and  in 
mice  (Marshak,  1936). 

Cytoplasmic  pollen  sterility  has  been  found  in  Oenothera,  Streptocarpus, 
Epilobium,  flax,  maize,  onions,  sugar  beets,  and  carrots.  In  every  case  that 
has  been  adequately  studied,  the  basic  sterility  remains  the  same  in  repeated 
generations  of  backcrossing,  but  the  amount  of  pollen  produced  varies  in 
different  genotypes.  There  is  an  interrelation  between  plasmagenes  and 
chromogenes  determining  the  final  result  (Jones,  1950). 

In  maize  the  amount  of  pollen  produced  ranges  from  0  to  100  per  cent. 
Only  by  suitable  tests  can  these  cases  of  full  fertility  be  recognized  as  having 
any  cytoplasmic  basis.  Interest  in  this  problem  now  centers  on  the  effect  of 
these  cytoplasmic  differences  on  heterosis. 

A  series  of  standard  inbreds  have  been  converted  by  crossing  these  onto 
suitable  sterilizer  stocks,  and  backcrossing  a  sufficient  number  of  generations 
to  re-establish  completely  the  inbred,  and  maintaining  the  inbred  in  a  sterile 
condition  by  continuous  backcrossing.  It  has  been  found  necessary  to  select 
both  the  cytoplasmic  sterile  seed  parent  individuals  and  the  individual  fertile 
pollen  parents  for  their  ability  to  maintain  complete  sterility  both  in  inbreds 
and  in  crosses.  In  some  lines  it  has  proved  to  be  impossible  to  establish  com- 
plete sterility,  but  the  majority  are  easily  sterilized  and  maintained  in  that 
condition. 

A  comparison  of  fertile  and  sterile  progenies  in  inbreds,  in  single  crosses  of 
two  inbreds,  and  multiple  crosses  of  three  and  four  inbreds,  shows  that  this 


234 


DONALD  F.  JONES 


cytoplasmic  difference  has  no  appreciable  effect  on  size  of  plant  as  measured 
by  height  at  the  end  of  the  season,  in  days  to  silking,  or  in  yield  of  grain. 
The  results  are  given  in  Table  14.3.  With  respect  to  pollen  sterility-fertility, 
the  cytoplasm  has  no  effect  on  heterosis. 

In  the  conversion  of  standard  inbreds  to  the  cytoplasmic  sterile  pollen 
condition,  it  has  been  found  that  many  of  these  long  inbred  strains,  presum- 
ably highly  homozygous,  are  segregating  for  chromogenes  that  have  the  abil- 
ity to  restore  pollen  fertility.  In  normally  fertile  plants  these  genes  have  no 
way  of  expressing  themselves.  They  are  not  selected  for  or  against  unless 
they  contribute  in  some  way  to  normal  pollen  production.  It  is  one  more 

T.\BLE  14.3 
A  COMPARISON  OF  FERTILE  AND  STERILE  MAIZE  PLANTS 


Fertile 

Sterile 

5  Inbreds 

7  Crosses  of  two  inbreds. . 
7  Crosses  of  two  inbreds     . 
3  Crosses  of  three  inbreds. 

1  Cross  of  three  inbreds.  .  . 
3  Crosses  of  four  inbreds 

5  Crosses  of  four  inbreds.  . 

2  Crosses  of  four  inbreds    . 

72.3 
102.6 

58.5 
111.7 

99.1 
123.9 

61.1 
115.8 

70.1     Height  of  Stalk 

97.7     Height  of  Stalk 

58.3     Days  to  first  silk 
108.9     Yield,  bushels  per  acre 
103.3     Yield,  bushels  per  acre 
119.0     Yield,  bushels  per  acre 

64.5     Yield,  bushels  per  acre 
117.3     Yield,  bushels  per  acre 

14  Crosses,  average  yield.  . .  . 

102.8 

102 . 6     Yield,  bushels  per  acre 

source  of  evidence  to  show  that  there  is  a  considerable  amount  of  enforced 
heterozygosity  in  maize.  Even  highly  inbred  families  remain  heterozygous. 
This  has  been  shown  to  be  true  for  other  species  of  plants  and  animals. 

SUMMARY 

Specific  evidence  from  a  study  of  chlorophyll  production  in  maize  and 
from  similar  studies  in  Neurospora,  Drosophila,  and  other  plant  and  animal 
species  proves  conclusively  that  there  are  numerous  mutant  genes  that  re- 
duce the  ability  of  the  organism  to  grow  and  to  survive.  Such  genes  exist 
in  naturally  self-fertilized  and  cross-fertilized  organisms  and  in  arti- 
ficially inbred  families  such  as  maize.  The  normal  alleles  of  these  mutant 
genes  show  either  complete  or  partial  dominance,  and  any  crossbred  indi- 
vidual contains  a  larger  number  of  these  dominant,  favorable  alleles  than 
any  inbred  individual. 

Evidence  from  Nicotiana  shows  that  there  is  an  interaction  between  di- 
vergent alleles  at  the  same  locus  such  that  the  heterozygote  produces  a  larger 
amount  of  growth  and  a  higher  reproductive  rate  than  either  homozygote. 
There  is  no  increase  in  growth  rate  and  this  instance  is  considered  not  to  be 
heterosis.  The  assumption  of  an  increased  growth  rate,  or  true  heterosis,  in 
such  allelic  interactions  is  not  supported  by  specific  evidence  that  cannot  be 


PLASMAGENES  AND  CHROMOGENES  IN  HETEROSIS  235 

interpreted  in  other  ways.  The  experimental  evidence  to  date  does  not  dis- 
tinguish clearly  between  a  general  physiological  interaction  and  a  specific 
contribution  from  favorable  dominant  effects.  More  evidence  on  this  point  is 
needed. 

Backcrossing  experiments  in  maize,  where  dominant  gene  markers  are 
maintained  in  a  heterozygous  condition,  show  heterosis  continuing  to  the 
sixth  generation.  The  approach  to  the  level  of  growth  activity  of  the  recur- 
rent inbred  parent  is  so  slow  as  to  indicate  that  every  region  of  the  chromo- 
somes, divisible  by  crossing  over,  has  an  effect  on  growth. 

The  growth  rate  in  these  backcrossed  generations  is  maintained  at  a  level 
appreciably  above  the  proportional  number  of  heterozygous  allelic  pairs. 
This  effect  can  be  interpreted  in  a  number  of  ways  other  than  a  general 
physiological  interaction,  such  as  enforced  heterozygosity,  and  the  comple- 
mentary action  of  dominant  genes  at  different  loci. 

There  is  no  way  known  at  the  present  time  to  distinguish  clearly  between 
the  accumulation  of  favorable  dominant  effects  of  compound  or  multiple 
genes  at  the  same  loci  and  a  general  physiological  interaction  or  over- 
dominance. 

Reciprocal  crosses  differ  in  many  species,  resulting  in  appreciable  diver- 
gence in  the  amount  of  growth,  and  these  differences  have  a  cytoplasmic  basis. 
The  evidence  from  maize,  however,  shows  clearly  that  cytoplasmic  pollen 
sterility  has  no  effect  on  size  of  plant,  time  of  flowering,  or  productiveness. 


M.  R.  IRWIN 

Universify  of  Wisconsin 


Chapter   15 

Specificity 
of  Gene  Effects 


If  an  attempt  were  made  to  survey  all  the  possible  ramifications  suggested 
by  the  title  of  this  paper,  it  should  include  much  of  the  published  work  in 
genetics.  It  is  of  course  a  truism  to  all  students  of  genetics  to  state  that  some 
sort  of  differential  specificity  towards  the  end  product  must  exist  between 
allelic  genes  or  their  effects  could  not  be  studied.  It  would  be  very  interesting 
as  a  part  of  this  discussion  to  attempt  to  trace  the  change  in  concepts  held  by 
various  workers  during  these  past  fifty  years  concerning  the  nature  and  paths 
of  action  of  the  gene.  However,  beyond  a  few  remarks,  such  considerations 
are  hardly  within  the  scope  of  this  chapter. 

Since  the  effects  of  genes  can  be  recognized  only  if  there  are  differences  in 
the  end  product,  it  is  quite  natural  that  the  differences  in  the  experimental 
material  first  subjected  to  genetic  analyses  should  have  been  those  which 
were  visible,  as  differences  in  form,  color,  etc.  Although  the  pendulum  has 
swung  somewhat  away  from  intensive  investigations  of  such  hereditary  char- 
acteristics, it  should  be  emphasized  that  by  their  use  the  underlying  mecha- 
nisms of  heredity  have  been  elucidated. 

Major  attention  was  given  by  most  of  the  investigators  during  the  first 
quarter  of  this  century  to  the  effects  of  respective  genes  upon  individual 
hereditary  characters.  In  some  quarters  there  was  an  oversimplification  in 
the  interpretation  of  the  relation  of  the  gene  to  the  character  affected  by  it. 
Gradually,  however,  the  concept  has  become  clearer  that  the  majority  of 
hereditary  characters — even  many  of  those  which  had  previously  appeared 
to  be  most  simply  inherited — are  affected  by  many  genes. 

An  early  observation  of  gene  specificity,  too  long  neglected  by  all  but  a  few 
geneticists,  was  that  made  by  Garrod  in  1909  (see  1923  edition  of  Inborn 

*  Paper  No.  433  from  the  Department  of  Genetics,  University  of  Wisconsin. 

236 


SPECIFICITY  OF  GENE  EFFECTS  237 

Errors  of  Metabolism)  on  the  inability  of  some  humans  to  break  down  homo- 
gentisic  acid  (2,5-dihydroxy])henylacetic  acid),  resulting  in  the  disease  known 
as  alcaptonuria.  Observations  reported  by  Gross  (1914)  indicated  that  this 
affliction  was  due  to  the  lack  of  a  ferment  (enzyme)  in  the  serum  of  alcapto- 
nurics,  whereas  the  enzyme  capable  of  catalyzing  the  breakdown  of  homo- 
gentisic  acid  was  demonstrable  in  the  serum  of  normal  individuals.  As 
Beadle  (1945)  has  stated,  no  clearer  example  exists  today  that  "a  single  gene 
substitution  results  in  the  absence  or  inactivity  of  a  specific  enzyme  and  that 
this  in  turn  leads  to  the  failure  of  a  particular  biochemical  reaction."  (The 
writer  distinctly  remembers  that,  while  he  was  a  student  in  a  class  in  physio- 
logical chemistry,  the  instructor  paid  considerable  attention  to  the  chemical 
explanation  of  alcaptonuria,  but  none  at  all  to  its  hereditary  nature.) 

Another  example  of  gene  specificity  and  also  of  gene  dosage  is  that  of  yel- 
low endosperm  in  corn  and  the  content  of  vitamin  A  reported  by  Mangelsdorf 
and  Fraps  (1931).  Their  study  showed  that  the  amount  of  vitamin  A  in  the 
endosperm  of  white  corn  was  almost  negligible,  but  that  the  presence  in  the 
endosperm  of  one,  two,  or  three  genes  for  yellow  pigmentation  was  accom- 
panied by  corresponding  increases  in  the  amounts  of  the  vitamin. 

GENE  EFFECTS  IN  A  SERIES  OF  REACTIONS 

There  are  numerous  examples  which  have  shown  that  many  genes  con- 
tribute to  the  development  of  a  heritable  character.  Thus,  in  corn  there  are 
many  genes  which  affect  the  development  of  chlorophyll.  Each  recessive 
allele,  when  homozygous,  allows  the  formation  of  only  partial  pigmentation, 
or  in  extreme  cases  no  pigmentation  at  all,  and  the  seedlings  are  albino.  It  is 
generally  believed  that  the  majority,  if  not  all,  of  these  different  genes  for 
albinism  affect  different  steps  in  the  process  of  chlorophyll  development.  A 
breakdown  of  the  process  at  any  one  of  these  steps  results  in  albinism  of  the 
seedling.  Haldane  (1942)  has  likened  the  complexity  of  such  a  synthetic  proc- 
ess to  the  activity  of  an  equal  number  of  students  as  there  are  genes,  "en- 
gaged on  different  stages  of  a  complicated  synthesis  under  the  direction  of  a 
professor,  except  that  attempts  to  locate  the  professor  have  so  far  failed.  Or 
we  may  compare  them  to  modern  workers  on  a  conveyor  belt,  rather  than 
skilled  craftsmen  each  of  whom  produces  a  finished  article." 

One  of  the  earliest  examples  of  the  physiological  bases  of  the  specificities 
which  are  the  final  gene  products  is  that  of  the  chemical  analyses  of  genetic 
variations  in  flower  color.  These  studies  were  carried  out  in  England  by  sev- 
eral workers.  See  reviews  by  Beadle  (1945),  Beale  (1941),  Haldane  (1942), 
Lawrence  and  Price  (1940)  for  the  general  results  and  references  to  specific 
papers. 

Mention  will  be  made  here  of  only  one  of  the  many  investigations  which 
have  defined  in  chemical  terms  the  hereditary  differences  in  pigmentation. 
Anthocyanin  is  one  of  the  five  types  of  pigments  concerned  in  flower  color, 


238  M.  R.  IRWIN 

and  its  presence  or  absence  in  several  species  is  genetically  determined.  One 
way  in  which  anthocyanin  may  be  modified  is  by  the  degree  of  oxidation  of 
the  prime  ring.  According  to  Beale  (1941)  in  the  two  genera  Lathyrus  and 
Streptocarpus,  the  hydroxyl  group  is  at  position  4'  in  the  pelargonidin  type, 
at  positions  3'  and  4'  in  the  cyanidin  types,  and  at  3',  4',  and  5'  in  the  delphin- 
idin  types.  The  more  oxidized  pigments  are  usually  dominant  to  the  less 
oxidized  types.  Thus  flowers  with  genes  AB  and  -46  will  be  of  the  delphinidin 
type  of  pigment,  those  with  aB  of  the  cyanidin  type,  and  those  with  ab  of  the 
pelargonidin  type. 

These  and  other  extensive  chemical  studies  on  the  anthocyanin  pigments 
genetically  modified  in  various  ways  are  dramatic  examples  of  the  specifici- 
ties of  gene  effects.  The  analogy  drawn  above  between  the  various  genes  and 
students  working  on  a  complicated  synthesis  becomes  a  little  more  clear  in 
relation  to  flower  pigments,  since  considerable  information  is  available  as  to 
what  some  of  the  genes  accomplish. 

A  further  example  of  the  effect  of  many  genes  upon  a  character  is  that  of 
eye  color  in  Drosophila  melanogaster.  Between  twenty-five  and  thirty  genes 
are  known  to  modify  the  brownish-red  color  of  the  wild-type  eye.  There  ap- 
pear to  be  two  independent  pigments,  brown  and  red,  concerned  in  the  devel- 
opment of  the  wild-type  eye,  each  of  these  being  affected  by  specific  genes. 
Certain  components  of  the  brown  pigment  are  diffusible  from  one  part  of  the 
body  to  another,  and  hence  are  more  readily  subjected  than  others  to  chem- 
ical analyses. 

The  details  of  these  analyses  are  presented  in  other  review  articles  (Beadle, 
1945;  Ephrussi,  1942a,  1942b).  Briefly,  dietary  tryptophan  is  converted  to 
alpha-oxytryptophan  by  a  reaction  controlled  by  the  wild-type  allele  of  the 
vermilion  gene  (?).  This  substance  is  oxidized  further  to  kynurenine  (the  so- 
called  i'+  substance).  By  virtue  of  the  activity  of  the  normal  allele  of  the  cin- 
nabar gene,  kynurenine  is  further  oxidized  to  the  or  substance,  which  may 
be  the  chromogen  of  the  brown  pigment  (Kikkawa,  1941).  The  production  of 
either  brown  or  red  eye  pigment  can  be  blocked  by  genes  at  the  white  eye 
locus,  thus  indicating  that  such  genes  act  on  a  common  precursor  of  the  red 
and  brown  pigments. 

Mention  should  be  made  of  the  relation  between  the  original  designation 
of  certain  of  the  genes  for  eye  color  and  their  presently  known  effects.  Thus, 
the  eyes  of  flies  with  the  mutant  alleles  bw  bw  are  brown.  But  it  is  now  known 
that  this  pair  of  alleles,  instead  of  being  concerned  with  the  production  of 
brown  pigment,  restricts  the  development  of  red  pigment  and  thus  we  see 
only  the  brown  color.  Similarly,  the  four  gene  pairs  whose  mutants  modify 
the  red  coloration  do  so  by  virtue  of  their  effect  on  the  brown  pigment,  not 
upon  the  red. 

Wheldale  (1910)  proposed  four  decades  ago  that  genetic  characters  were 
the  resultant  of  a  series  of  reactions,  and  that  if  a  break  in  the  chain  occurred, 


SPECIFICITY  OF  GENE  EFFECTS  239 

the  series  of  steps  would  have  proceeded  only  to  that  j)oint.  Following  the 
initial  work  in  Neurospora  by  Beadle  and  Tatum  (1941)  on  mutants  which 
blocked  certain  metabolic  processes,  this  type  of  approach  has  expanded 
enormously  and  profitably.  Attention  can  be  called  here  to  but  one  very  sig- 
nificant example  of  this  kind  of  experimental  study  in  microorganisms.  A 
report  by  Srb  and  Horowitz  (1944)  shows  clearly  how  many  genes  act  in  the 
synthesis  of  arginine.  Of  fifteen  mutant  strains  studied,  there  were  seven  dif- 
ferent steps  represented  in  the  synthesis  of  arginine.  One  of  the  forms  grew 
only  if  arginine  was  supplied.  Two  others  required  either  arginine  or  citrul- 
line,  and  these  two  strains  were  genetically  different.  Four  other  strains, 
genetically  different  from  the  first  three  strains  and  from  each  other,  would 
grow  if  arginine,  citrulline,  or  ornithine  were  provided.  For  a  diagrammatic 
representation  of  these  steps,  see  Beadle  (1945). 

DIRECT  EFFECTS  OF  GENES 

The  preceding  examples  are  but  a  few^  of  the  many  which  could  be  cited  to 
illustrate  the  gene  specificities  in  the  development  of  a  genetic  character 
which  involves  the  successive  activities  of  many  genes.  Are  there  any  genetic 
characters  which  may  be  the  immediate  products  of  the  causative  genes?  An 
example  almost  unique  in  higher  plants  is  that  of  the  waxy  gene  in  corn 
(Collins,  19C9)  in  its  effects  upon  the  starch  of  the  pollen  grain  and  the  endo- 
sperm reserves.  As  is  well  known,  the  starch  granules  in  the  pollen  grains 
bearing  the  waxy  gene  are  stained  reddish-brown  with  iodine,  as  are  the 
endosperm  reserves  of  waxy  seeds,  in  contrast  to  the  typical  blue  reaction  of 
the  starch  granules  of  non-waxy  pollen  and  of  the  endosperm  reserves  of 
non-waxy  seeds.  Following  studies  of  the  physiological  effects  of  the  waxy 
gene,  Brink  (1929)  proposed  that  this  gene  has  its  effect  on  the  enzyme  amy- 
lase which  functions  directly  in  the  synthesis  of  starch. 

Another  class  of  hereditary  characters  which  in  some  respects  appears  to 
satisfy  some  of  the  criteria  for  a  direct  effect  of  the  causative  genes  is  that  of 
the  antigenic  characters  of  the  red  blood  cells  of  animals.  With  only  rare 
exceptions,  to  be  considered  later  in  more  detail,  each  of  the  known  antigenic 
substances  has  appeared  in  the  cells  of  an  individual  only  if  one  or  both 
parents  also  possessed  it.  If  there  is  but  a  single  pair  of  contrasting  charac- 
ters, each  is  expressed  in  the  heterozygote.  Further,  the  cells  which  give  rise 
to  the  hematopoietic  tissue  from  which  the  red  blood  corpuscles  are  derived 
are  laid  down  shortly  after  the  first  division  of  the  fertilized  egg.  The  possibil- 
ity cannot  be  excluded,  of  course,  that  there  is  a  chain  of  reactions  within 
each  cell  leading  to  the  formation  of  the  antigen,  but  no  block  in  such  a  chain 
of  reactions  has  yet  been  observed.  There  are  two  statements  concerning  the 
cellular  antigens  which  are  of  interest:  (1)  the  antigenic  substance  must  be 
located  at  or  near  the  surface  of  the  cell  in  order  to  be  detectable,  and  (2) 
there  is  no  known  effect  of  the  environment  upon  them. 


240  M.  R.  IRWIN 

We  should  avoid  misunderstanding  about  the  meaning  of  the  terms  com- 
monly used  in  immunological  literature.  For  example,  the  word  antigen  was 
originally  defined  as  any  substance  which,  when  introduced  parenterally  into 
an  animal,  would  invoke  the  production  of  antibodies.  This  definition  would 
now  be  extended  to  include  any  substance  which  will  react  visibly  with  an 
antibody.  And  an  antibody  would  be  defined  as  a  constituent  of  the  serum 
which  reacts  with  an  antigen  in  any  of  several  ways.  The  circle  of  reasoning 
here  is  obvious.  However,  insofar  as  chemical  studies  of  various  antigens  have 
contributed  to  an  understanding  of  their  specificities,  the  specificities  have 
always  been  associated  with  structural  differences  of  the  antigenic  sub- 
stances. On  the  other  hand,  the  reasons  underlying  the  specificities  in  re- 
activity of  the  antibodies  are  almost  completely  unknown,  although  it  is 
known  that  the  antibodies  are  intimately  associated  with  the  globulins  of  the 
serum,  and  in  fact  may  constitute  the  gamma  globulins  of  the  serum. 

CELLULAR  ANTIGENS  IN  HUMANS 

As  our  first  example  of  these  antigenic  substances,  let  us  consider  the  well 
known  and  extensively  studied  O,  A,  B,  and  AB  antigenic  characters,  or 
blood  groups,  of  human  cells.  Following  their  discovery  by  Landsteiner 
(1900, 1901),  it  soon  became  clear  that  these  substances  were  gene  controlled. 
At  the  present  time,  the  theory  of  three  allelic  genes,  as  postulated  by  Bern- 
stein (1924)  on  statistical  grounds,  is  generally  accepted.  The  two  other  the- 
ories proposed  for  their  inheritance — independent  and  linked  genes,  respec- 
tively— are  fully  discussed  by  Wiener  (1943).  Landsteiner  noted  that  the 
serum  of  certain  individuals  would  agglutinate  (clump)  the  cells  of  other 
individuals,  and  from  this  observation  the  reciprocal  relationship  between  the 
presence  and  absence  of  each  antigen  and  its  specific  antibody  has  been 
elucidated. 

A  or  B  Antibody 

Antigen  on  of  the 

Group  the  Cells  Serum 

O None  Anti-A,  Anti-B 

A A  Anti-B 

B B  Anti-A 

AB AB  None 

It  may  readily  be  seen  that  the  presence  of  an  antigen,  as  A,  on  the  cells 
is  accompanied  by  the  presence  of  the  antibody  (anti-B)  for  the  contrasting 
antigen,  as  B,  in  the  serum,  and  vice  versa.  If  both  antigenic  characters  are 
found  on  the  cells,  as  in  AB  individuals,  the  serum  contains  no  antibodies. 
While  if  neither  A  nor  B  is  present  on  the  cells,  the  serum  contains  both  anti- 
A  and  anti-B. 

These  phenomena  pose  the  question  whether  the  genes  producing  the  cel- 
lular substances  also  have  an  effect  on  the  antibodies  of  the  serum.  That  is, 
does  the  gene  which  is  responsible  for  the  O  antigen  (which  is  definitely  an 
entity  but  is  less  reactive  than  A  and  B)  also  effect  both  anti-A  and  anti-B 


SPECIFICITY  OF  GENE  EFFECTS  241 

in  the  serum — while  in  individuals  with  substance  A,  only  anti-B  is  found; 
in  those  with  B,  only  anti-A  is  present;  and  in  AB  individuals  the  effects  of 
the  respective  genes  on  the  antibodies  are  somehow  neutralized? 

Before  attempting  to  answer  this  question,  it  will  be  advisable  to  review 
the  present  knowledge  of  the  chemistry  of  the  A  and  B  substances  of  human 
cells.  See  Rabat  (1949). 

These  antigens  (blood  groups,  cellular  characters,  antigenic  factors,  etc.) 
are  found  in  nearly  all  the  fluids  and  tissues  of  the  human  body.  They  also 
are  widely  distributed  throughout  the  animal  kingdom.  The  A  substance  or 
an  A-like  substance  has  been  found,  for  example,  in  hog  gastric  mucosa,  in 
the  fourth  stomach  (abomasum)  of  the  cow,  and  in  swine  pepsin,  while  both 
A  and  B  substances  have  been  noted  in  the  saliva  and  stomachs  of  horses. 
Following  chemical  fractionations,  principally  of  horse  saliva  and  hog  gastric 
mucosa,  various  investigators  have  obtained  preparations  with  activity  re- 
lated to  the  A  substance.  These  preparations  have  been  largely  polysac- 
charide in  nature.  In  addition  to  the  polysaccharides,  even  in  the  purest 
preparations,  some  workers  have  noted  traces  of  amino  acids. 

At  present,  while  it  appears  that  both  the  A  and  B  substances  of  human 
cells  may  be  classed  as  nitrogenous  polysaccharides,  no  information  is  avail- 
able as  to  the  structural  differences  between  them.  Our  knowledge  of  such 
specificities  rests  entirely  upon  the  technics  of  immunology,  that  is,  by  the 
interaction  of  either  naturally  occurring  antibodies  (as  anti-A  and  anti-B), 
or  immune  antibodies,  with  the  respective  substances  A  and  B. 

The  antigenic  substances  A  and  B  of  human  cells  are  complex  polysac- 
charides, while  the  antibodies  are  modified  globulins,  or  are  found  in  serum 
protein  very  closely  related  to  the  globulin  fraction.  If  the  gene  which  effects 
antigen  A  is  responsible  also  for  the  B  antibody,  and  that  for  antigen  B  for  the 
A  antibody,  it  would  seem  that  here  is  a  clear-cut  case  of  pleiotropic  effects  of 
the  respective  genes.  This  explanation  runs  into  difficulties  in  AB  individuals 
which,  on  this  proposal,  should  have  both  kinds  of  antibodies  but  actually 
have  none.  In  contrast,  a  current  explanation  of  the  reciprocal  presence  of  the 
antigenic  substance  of  the  cells  and  the  antibody  for  the  contrasting  sub- 
stances is  that  the  antibodies  for  both  substances  (A  and  B)  are  normal  con- 
stituents of  human  serum.  Production  of  the  antibodies  would  then  be  con- 
trolled by  a  gene  or  genes  at  another  locus  than  that  having  to  do  with  the 
cellular  substances,  ij  genes  were  involved  in  their  production.  If  an  indi- 
vidual carries  the  gene  for  A,  and  hence  has  A  substances  widely  distributed 
throughout  his  body,  the  A  antibodies  are  presumed  to  be  absorbed  from  the 
serum,  and  of  course  the  B  antibodies  are  left.  Also,  an  individual  with  the 
B  substance  would  absorb  the  B  antibodies,  and  the  antibodies  to  A  would 
remain,  while  both  anti-A  and  anti-B  would  be  absorbed  in  an  AB  individual. 
Other  hypotheses  are  given  by  Wiener  (1943).  Unfortunately,  no  experimen- 
tal test  of  the  correctness  of  this  or  other  hypotheses  is  likely. 


242  M.  R.  IRWIN 

Landsteiner  and  Levine  (1927)  announced  the  discovery  in  human  cells  of 
a  new  pair  of  contrasting  antigens,  called  M  and  N.  These  were  detectable 
only  by  the  use  of  immune  sera  produced  in  rabbits,  as  was  another  antigenic 
factor  called  P.  The  heritability  of  the  M  and  N  substances  is  adequately 
explained  by  the  assumption  of  a  single  pair  of  allelic  genes,  and  the  sub- 
stance P  appears  to  be  dominant  to  its  absence. 

Another  antigenic  factor  in  human  blood  which  has  aroused  wide  interest 
is  the  recently  discovered  Rh  substance,  or  complex,  as  it  might  be  termed. 
In  1940,  Landsteiner  and  Wiener  (1940)  reported  that  a  new  antibody,  de- 
rived from  a  rabbit  immunized  with  the  erythrocytes  of  a  rhesus  monkey, 
was  reactive  with  the  cells  of  about  85  per  cent  of  the  white  population  of 
New  York.  They  gave  the  name  Rh  (a  contraction  of  rhesus)  to  this  agglu- 
tinable  property  of  human  cells.  As  Boyd  (1945)  aptly  states: 

The  technic  of  testing  for  the  new  factor  was  difficult,  the  best  available  serums  were 
weak,  and  had  it  not  been  for  a  remarkable  series  of  discoveries  which  followed  in  the  next 
few  months,  the  Rh  factor  might  have  aroused  no  more  interest  than  its  practically  still- 
born brethren. .  .  . 

The  Rh  factor  was  shown  to  be  involved  in  previously  unexplained  com- 
plications following  transfusions  (Wiener  and  Peters,  1940),  but  is  most 
widely  known  for  its  role  as  the  etiologic  agent  in  the  majority  of  cases 
of  hemolytic  disease  of  the  newborn.  The  proposal  was  first  made  by 
Levine  and  Stetson  (1939)  that  an  antigen  in  the  fetus,  foreign  to  the  mother 
and  presumably  transmitted  by  the  father,  could  pass  through  the  placenta 
and  immunize  the  mother.  Later  studies  implicated  the  Rh  factor  as  the 
foreign  antigen,  and  showed  that  the  antibodies  developed  in  the  mother  may 
pass  back  through  the  placenta  and  affect  the  red  blood  cells  of  the  fetus, 
before  or  following  birth.  Although  the  majority  of  cases  of  hemolytic  disease 
of  the  newborn  may  be  justly  ascribed  to  Rh  incompatibility  between  the 
father  and  mother,  there  is  no  satisfactory  explanation  as  to  why  only  about 
one  in  forty  of  such  potentially  dangerous  combinations  leads  to  morbidity. 

There  exist  several  subgroups,  or  subtypes,  of  the  Rh  complex,  and  inves- 
tigations as  to  their  respective  specificities  occupy  the  center  of  interest  of 
many  workers  at  the  present  writing.  There  are  two  schools  of  thought  as  to 
the  mode  of  inheritance  of  these  subgroups,  which  also  involves  the  terminol- 
ogy to  be  used  in  their  identification  (see  Strandskov,  1948,  1949,  for  leading 
references).  One  explanation  is  that  the  various  subtypes  are  manifestations 
of  a  series  of  multiple  allelic  genes,  the  other  that  they  are  the  result  of  the 
action  of  respective  genes  at  three  different  but  closely  linked  loci.  It  is  not 
within  the  province  of  this  chapter  to  discuss  the  arguments  for  and  against 
these  two  proposals.  However,  it  should  be  stated  that  the  genetic  results 
under  either  explanation  are  essentially  the  same. 

One  of  the  most  pertinent  statements  which  can  be  made  about  these 
various  antigenic  substances  of  the  erythrocytes  is  that  they  are  detectable 


SPECIFICITY  OF  GENE  EFFECTS  243 

no  matter  in  what  gene  complex  they  may  occur.  That  is,  other  genes  than 
the  causative  ones  have  no  measurable  influence  upon  their  expression.  A 
possible  exception  to  this  statement  might  be  proposed  for  the  A  and  N  char- 
acters, res])ectively,  since  each  is  somewhat  less  readily  agglutinated  when  in 
the  heterozygote,  AB  and  MN,  than  when  either  occurs  singly. 

THE  HYBRID  SUBSTANCE  IN  SPECIES  HYBRIDS 
Until  the  early  part  of  this  century,  most  of  the  workers  in  immunology 
had  reached  the  conclusion  that  the  specificities  obtained  in  immunological 
reactions  were  primarily  if  not  entirely  concerned  with  proteins.  Therefore, 
the  finding  by  Heidelberger  and  Avery  (1923,  1924)  that  the  immunological 
specificities  of  the  pneumococcal  types  were  dependent  upon  polysaccharides 
was  indeed  a  forward  step  in  our  understanding  of  the  chemical  nature  of 
biological  specificity.  It  is  a  pleasure  to  acknowledge  that  this  work  of 
Heidelberger  and  Avery  convinced  the  writer  that  immunological  technics 
should  be  a  useful  tool  in  studying  genetic  phenomena.  Also,  although  at  that 
time  pollen  differing  in  gene  content  seemed  (and  still  does)  to  be  promising 
experimental  material,  the  species  and  species  hybrids  in  pigeons  and  doves 
produced  by  the  late  L.  J.  Cole  were  tailor-made  for  further  studies. 

Pigeon-Dove   Hybrids 

The  first  step  was  to  determine  whether  the  cells  of  one  species  could  be 
distinguished  from  those  of  the  other.  In  brief,  all  the  comparisons  by  im- 
munological technics,  between  any  pair  of  species  of  pigeons  and  doves,  have 
resulted  in  the  ability  to  distinguish  the  cells  of  any  species  from  those  of  an- 
other, and  to  show  that  each  species  possessed  antigenic  substances  in  com- 
mon with  another  species,  as  well  as  those  peculiar  to  itself — those  species 
specific.  A  dozen  or  more  kinds  of  species  hybrids  have  been  obtained  in  the 
laboratory,  and  in  general,  each  kind  of  hybrid  has  contained  in  its  cells  all 
or  nearly  all  of  the  cellular  substances  of  both  parental  species.  One  such 
species  hybrid  is  that  obtained  from  a  mating  between  males  of  an  Asiatic 
species,  the  Pearlneck  (Slreplopelia  chinensis)  and  the  domesticated  Ring 
dove  females  {St.  risoria).  The  corpuscles  of  these  hybrids  contained  all  the 
substances  common  to  each  parental  species,  but  did  not  contain  quite  all  the 
specific  substances  of  either  parental  species.  Further,  the  cells  of  these 
hybrids  did  possess  a  complex  of  antigenic  substances  not  found  in  the  cells 
of  the  parents.  These  relationships  are  presented  in  Table  15.1  and  are  given 
in  diagrammatic  form  in  Figure  15.1.  This  new  antigen  has  been  called  the 
"hybrid  substance,"  and  it  has  been  ])resent  in  every  hybrid  produced  be- 
tween these  two  species. 

Upon  repeatedly  backcrossing  these  species  hybrids  and  selected  back- 
cross  hybrids  to  Ring  dove,  ten  antigenic  substances  which  differentiate 
Pearlneck  from  Ring  dove  have  been  isolated  as  probable  units.  That  is,  a 


244 


M.  R.  IRWIN 


backcross  bird  carrying  any  one  of  these  unit  substances,  when  mated  to  a 
Ring  dove,  has  produced  approximately  equal  proportions  of  progeny  with, 
and  without,  the  particular  substance  in  their  blood  cells.  These  substances 
peculiar  to  Pearlneck,  as  compared  with  Ring  dove,  have  been  called  d-1, 
d-2,  d-3,  d-4,  d-5,  d-6,  d-7,  d-8,  d-11,  and  d-12.  Each  of  these  is  distinct  from 
the  others  (Irwin,  1939)  both  genetically  and  immunologically.  Thus  it  ap- 
pears that  a  gene  or  genes  on  each  of  ten  of  the  thirty-odd  pairs  of  chromo- 

TABLE  15.1 

ANTIGENIC  RELATIONSHIPS  OF  THE  BLOOD  CELLS 

OF  PEARLNECK,  RING  DOVES,  AND 

THEIR  HYBRIDS 


Immune  Serum 


Pearlneck . 
Pearlneck . 
Pearlneck . 


Ring  dove. 
Ring  dove. 
Ring  dove . 


Fi. 
Fi. 
F,. 
Fi. 


Absorbed  by 
Cells  of 


Ring  dove 
Fi 


Pearlneck 
Fi 


Pearlneck 
Ring  dove 
Pearlneck  and 
Ring  dove 


Agglutination  Titers 
WITH  Cells  of 


Pearlneck 


23040 

11520 

90 

15360 
0 
0 

15360 

0 

7680 

0 


Ring  Dove 


23040 
0 
0 

15360 

3840+ 
180 

15360 
3840+ 
0 
0 


23040 

11520 

0 

15360 
3840+ 
0 

15360 
3840+ 
7680 
360+ 


"  -o  ^'^  o"q  a  o 


.<>%»o  •  Pearlneck  .  <,  o.-/ 

On     a  oQf,    a    n     M     /i    1%    n       ^    O    O  O    O   «     O 


I 


A  a.  H.Q a  a  q  q  • 


I 


t\\ Pearlneck  "aV-sfCs; 


Common 


Common 

Ringdove 

1 

i 

Pearlneck 


Ringdove 


Common- 


Ringdove^  P'N.  x  R,D.-F, 


Hybrid 


Fig.  15.1 — Diagrammatic  representation  of  the  antigenic  relationships  of  the  Pearlneck, 

Ring  dove,  and  their  hybrids. 


SPECIFICITY  OF  GENE  EFFECTS  245 

somes  of  Pearlneck  produce  effects  on  cellular  antigens  which  differentiate 
Pearlneck  from  Ring  dove.  Although  the  cellular  substances  particular  to 
Ring  dove,  in  contrast  to  Pearlneck,  have  not  been  obtained  as  units,  the 
available  evidence  indicates  strongly  that  a  gene  or  genes  on  nine  or  ten 
chromosomes  of  Ring  dove  produce  antigenic  effects  which  differentiate  that 
species  from  Pearlneck. 

The  question  may  well  be  raised  as  to  what  this  recital  of  antigenic  char- 
acters in  man  and  doves,  which  in  general  illustrates  gene  specificity  in  the 
production  of  cellular  antigens,  has  to  do  with  the  general  topic  of  heterosis. 
The  so-called  hybrid  substance  has  one  word  (hybrid)  in  common  with  the 
term  hybrid  vigor,  and  suggests  a  possible  relationship  of  the  two  terms. 

The  hybrid  substance  seemingly  represents  a  departure  from  the  hypoth- 
esized direct  action  of  a  gene  on  the  antigenic  substance,  in  that  it  appears 
to  result  from  the  interaction  of  two  or  more  genes  in  the  species  hybrids  to 
produce  some  antigenic  substance  different  from  any  detectable  in  either 
parent.  With  but  one  exception  proposed  by  Thomsen  (1936)  in  chickens,  and 
for  which  another  explanation  will  be  considered  shortly,  a  hybrid  substance 
has  thus  far  been  found  only  in  species  hybrids. 

Mention  should  be  made  of  the  technics  required  for  the  detection  of  the 
hybrid  substance.  Briefly,  if  an  antiserum  prepared  against  the  cells  of  an 
individual,  whether  a  species  hybrid  or  not,  would  be  absorbed  by  the  cells  of 
both  its  parents  and  would  then  react  with  the  cells  of  the  individual,  but  not 
with  the  cells  of  either  parent,  there  would  be  evidence  of  a  different  anti- 
genic substance  in  the  homologous  cells — those  used  in  the  immunization. 
(If  an  antigen  were  recessive,  it  would  be  present  in  the  heterozygote,  and 
presumably  could  absorb  its  specific  antibody.) 

Domestic    Fowl    Hybrids 

As  stated  above,  Thomsen  (1936)  reported  that  within  each  of  two  families 
of  chickens  there  was  a  different  antigenic  substance  present  than  was  found 
in  the  parents.  Attempts  in  our  laboratory  by  Mrs.  Ruth  Briles  to  duplicate 
this  finding  were  without  success,  but  a  very  interesting  and  quite  unex- 
pected observation  was  made  which  may  be  the  explanation  of  Thomsen 's 
finding.  If  an  antigenic  substance  were  present  in  an  individual  different  from 
that  possessed  by  either  parent,  immunization  of  either  parent  (as  #1)  with 
the  cells  of  this  individual  might  engender  antibodies  against  the  new  sub- 
stance. Absorption  of  such  an  antiserum  by  the  cells  of  the  other  parent  (as 
#2)  should  remove  all  antibodies  except  those  formed  against  the  new  or 
hybrid  substance,  and  such  a  reagent  should  be  reactive  only  with  the  cells 
containing  the  new  substance.  This  was  the  procedure  followed  by  Thomsen, 
except  that  his  tables  do  not  show  that  the  cells  of  the  two  parents  were  used 
as  negative  controls  in  the  tests  made  after  the  various  absorptions. 

Immunizations  of  each  of  the  parents  of  a  family  of  chickens  against  the 


246  M.  R.  IRWIN 

cells  of  one  of  the  offspring,  or  the  pooled  cells  of  two  or  more,  were  made  by 
Mrs.  Briles.  Following  the  absorption  of  the  antiserum  obtained  from  either 
parent  by  the  cells  of  the  other,  it  was  noted  that  the  absorbed  antiserum  was 
at  least  weakly  reactive  with  the  cells  of  the  individual  from  which  the  anti- 
serum was  obtained.  That  is,  such  an  antiserum  would  not  react  (agglutinate) 
with  its  own  cells  before  absorption  with  the  cells  of  the  mate,  but  after  such 
absorption  it  definitely  would  agglutinate  the  cells  of  the  individual  from 
which  it  was  derived. 

To  use  a  concrete  example,  bird  R614  (containing  Bi  antigen)  was  im- 
munized with  the  washed  cells  of  R2C43,  to  produce  B3  antibodies  (Briles, 
McGibbon,  and  Irwin,  1951).  After  this  antiserum  from  R614  was  mixed  for 
absorption  with  the  washed  cells  of  R622  (having  B3  antigen  in  its  cells  and 
having  been  immunized  to  produce  Bi  antibodies),  all  cells  containing  the  Bi 
antigen  were  reactive  with  it,  including  those  of  R614  itself.  Thus  it  appears 
that  the  antibodies  to  Bi  which  were  circulating  in  the  serum  of  R622  were 
also  attached  to  the  surface  of  the  red  blood  cells  and  were  transferred  to  the 
antiserum  from  R614  during  the  absorption  process.  It  was  possible  to  dem- 
onstrate that,  after  washing  the  cells  of  R622  in  saline,  the  saline  contained 
antibodies,  even  after  nine  successive  washings.  Hence,  unless  the  cells  of 
both  parents  were  used  as  controls  in  comparable  tests  for  the  presence  of  a 
hybrid  substance,  agglutination  of  any  cells  could  be  explained  as  due  to  a 
transfer  of  antibody  from  the  blood  cells  to  an  antiserum.  Unfortunately, 
such  controls  are  not  given  in  Thomsen's  paper,  and  the  possibility  cannot  be 
eliminated  that  the  reactions  obtained  by  him  were  due  simply  to  segregation 
within  the  various  families  of  an  antigenic  character  of  one  of  the  parents. 
This  possibility  was  mentioned  by  Thomsen  (1936),  but  was  not  considered 
applicable  to  his  experiments. 

Hybrid  Substances 

Returning  to  the  hybrid  substance  for  which  there  is  definite  evidence,  it 
should  first  be  stated  that  such  a  substance  has  not  been  found  in  all  kinds  of 
species  hybrids,  as  may  be  seen  from  the  data  given  in  Table  15.2.  It  has  been 
reported  from  our  laboratory  in  hybrids  between  Pearlneck  and  Ring  dove, 
the  pigeon  {Columba  livia)  and  Ring  dove,  the  Mallard  {Anas  platyrhynchos) 
and  Muscovy  duck  (Cairina  moschata),  but  not  in  the  hybrids  between  the 
triangular  spotted  pigeon  (C.  guinea)  and  livia.  Irwin  (1947)  gives  the  spe- 
cific references  to  pertinent  articles. 

A  hybrid  substance  has  been  detected  but  not  previously  reported  in 
hybrids  from  malings  between  the  Philippine  turtle  dove  (St.  dussumieri) 
and  Ring  dove,  the  dwarf  turtle  dove  {St.  Jmmilis)  and  Ring  dove,  the 
Oriental  turtle  dove  {St.  orientalis)  and  Ring  dove,  and  the  band  tail  pigeon 
{C.  fasciata)  and  livia.  No  such  substance  has  been  observed  in  the  hybrids 
between  the  Senegal  dove  {St.  senegalensis)  and  Ring  dove,  an  African  dove 


SPECIFICITY  OF  GENE  EFFECTS 


247 


(St.  semitorquata)  and  Ring  dove,  the  Senegal  dove  and  the  Cape  turtle  dove 
{St.  capicola),  the  spot  wing  pigeon  (C.  maailosa)  and  livia,  and  between  the 
Grayson  dove  {Zenaidiira  graysoni)  and  the  common  mourning  dove  {Zen. 
macroura).  It  is  possible  that  a  hybrid  substance  does  exist  in  these  latter 
species  hybrids,  but  the  same  technics  by  which  it  was  observed  in  the  other 
species  hybrids  failed  to  demonstrate  its  presence  in  them. 

Three  different  fractions  of  the  hybrid  substance  have  been  demonstrated 
in  the  hybrids  between  Pearlneck  and  Ring  dove  (Irwin  and  Cumley,  1945), 
by  virtue  of  a  frequent  association  of  each  fraction  with  one  or  more  antigens 

TABLE  15.2 

TESTS  FOR  HYBRID  SUBSTANCES  IN  THE  CELLS  OF 

VARIOUS  SPECIES  HYBRIDS 


Antiserum  to 


Fi — Pearlneck  X  Ring  dove 

Fi — C.  //zi/aX  Ring  dove 

Fi — St.dussumieriXRing  dove. .  .  . 

Fi — St.  humilisXRing  dove 

Fi — St.  orientalisXRing  dove 

Fi — C.  fasciataXlivia 

Fi — Mallard  X  Muscovy 

Fi — St.  senegalensisXRing  dove.  .  . 
Fi — St.  semitorquataXRing  dove.  . 

F: — Senegal  X-S/.  capicola 

Fi — C.  tnaculosaXlivla 

Fi — C.  guineaXlivia 

Fi — Zenaidura    gray soniX  Zen.    ma 
crmira 


Absorbed  by  Cells  of 


Parent  1 


Parent  2 


Pearlneck 

livia 

dussumieri 

h  II  mil  is 

orientalis 

C.  fasciata 

Mallard 

St.  senegalensis 

St.  semitorquata 

Senegal 

C.  maculosa 

C.  guinea 

Zen.  graysoni 


Ring  dove 
Ring  dove 
Ring  dove 
Ring  dove 
Ring  dove 
livia 

Muscovy- 
Ring  dove 
Ring  dove 
St.  capicola 
livia 
livia 

TLen.  macroura 


Re.actions  of  Pa- 
rental AND  HYBRID 

Cells  with  the 

Respective 

Reagents 


Par- 

Par- 

ent 1 

ent  2 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

Hy- 
brid 


+  + 

+  + 

+ 

+ 

+ 

+ 

+  + 

0 

0 

0 

0 

0 

0 


peculiar  to  Pearlneck.  Thus  one  fraction  called  dx-A  was  always  associated 
in  the  backcross  hybrids  with  the  d-1 1  substance,  dx-B  seemingly  was  loosely 
linked  with  the  d-1  character  and  with  certain  others  as  well — thereby  pro- 
viding strong  evidence  that  on  several  chromosomes  of  Pearlneck  there  are 
duplicate  or  repeat  genes — and  dx-C  was  always  associated  with  the  d-4 
antigen.  The  pertinent  reactions  which  show  these  specificities  are  given  in 
Table  15.3  and  are  represented  diagrammatically  in  Figure  15.2. 

Because  of  the  constant  association  of  the  dx-A  and  dx-C  fractions  with 
the  d-11  and  d-4  substances,  respectively,  one  cannot  be  certain  that  these 
two  fractions,  although  antigenically  distinct  from  the  d-1  and  d-11  specific 
characters,  are  not  simply  a  new  specificity  conferred  upon  the  specific  char- 
acters by  some  sort  of  rearrangement  of  the  specific  substances  following  the 


248 


M.  R.  IRWIN 


interaction  of  the  causative  genes.  This  question  cannot  be  completely  an- 
swered until  either  a  genetic  separation  has  been  observed,  as  between  the 
dx-A  and  d-11,  or  the  chemical  separation  into  two  distinct  substances  has 
been  done.  On  the  other  hand,  the  dx-B  fraction  has  been  separated  from 
each  of  the  species  specific  characters  to  which  it  presumably  is  loosely 
linked,  thereby  showing  that  this  fraction  of  the  hybrid  substance  is  an 
antigenic  entity. 

The  reagent  which  interacts  with  the  hybrid  substance  (hybrid  antiserum 

TABLE  15.3 

TESTS  FOR  SIMILARITIES  AND  DIFFERENCES  OF  THE  COM- 
PONENTS OF  THE  "HYBRID  SUBSTANCE"  OF  THE  SPECIES 
HYBRID  BETWEEN  PEARLNECK  AND  RING  DOVE 


Reactions  of  Different  Cells  with  Anti-Fi  Serxjm 

Cells 

Absorbed 

by  Cells 

of  Both 

Pearlneck 

and  Ring 

Dove 

Absorbed  by  the  Cells  of  Both  Pearlneck  and  Ring  Dove, 
in  Combination  with  Others  as  Listed 

d-1 
(dx-B) 

d-4 

(dx-C) 

d-11 
(dx-A) 

P.N. 
^iR.D. 

Pgn 
^iR.D. 

Sen. 

Aus. 
cstd. 

Pearlneck 

Ring  dove 

Fi— P.N./R.D 

d-1  (dx-B). . . . 

d-4  (dx-C) .... 

d-11  (dx-A).  .  . 

Fi-Pgn/R.D 

Senegal 

Australian  crested. 

0 
0 

++ 

+ 

+ 

++ 

+ 

+ 

± 

0 
0 

++ 

0 

? 

++ 
+ 

0 

+ 

0 
0 

++ 
+ 

0 

++ 
+ 

0 

+ 

0 
0 

-1- 
+ 

-f 

0 

0 

0 

0 

0 
0 
0 
0 
0 
0 
0 
0 
0 

0 
0 

+  + 
+ 

-1- 

+ 
0 
0 
0 

0 
0 

++ 
+ 

-f- 

++ 
+ 

0 

± 

0 
0 

+  + 

+ 
-f 

+ 
-1- 
0 

Column 

2 

3 

4 

5 

6 

7 

8 

9 

Symbols:  -f -f  =  marked  agglutination;   -|-  =  agglutination;  ±   =  definite  but   weak  agglutination;   ? 
doubtful  reaction;  0  =  no  agglutination — at  the  first  dilution  of  the  serum  cell  mixture. 


absorbed  by  the  cells  of  both  Pearlneck  and  Ring  dove)  will  also  agglutinate 
the  cells  of  various  species.  Thus  in  the  genus  Streptopelia,  there  were  five 
species  {capicola,  dussutnieri,  kumilis,  orientalis,  and  senegalensis)  other  than 
Pearlneck  and  Ring  dove  whose  cells  were  reactive,  and  one  {semitorquata) 
with  nonreactive  cells.  Within  the  genus  Columba,  the  cells  of  one  species 
(rufina)  likewise  reacted  with  this  reagent,  but  those  of  seven  other  species 
(fasciala,flavirostris,  guinea,  livia,  maculosa,  palumbus,  and  picazura)  did  not. 
And  of  twelve  species  tested  in  other  genera  within  the  Columbidae,  only 
three  species  from  Australia  (Australian  crested  dove,  or  Ocyphaps  lopkofes, 
the  bronze  wing  dove  or  Phaps  chalcopiera,  and  the  brush  bronze  wing  dove, 
or  Phaps  elegans)  possessed  reactive  cells.  In  Table  15.3,  the  Senegal  cells  are 
representative  of  the  parallel  reactions  of  the  five  species  of  the  Streptopelia, 


SPECIFICITY  OF  GENE  EFFECTS 


249 


as  are  those  of  the  Australian  crested  of  the  equivalent  reactivities  of  the 
three  Australian  species. 

Although  the  reagent  for  the  hybrid  substance  did  not  agglutinate  the 
cells  of  livia,  it  invariably  clumped  those  of  the  hybrids  between  pigeon 
(livia)  and  Ring  dove.  As  previously  reported  (Irwin  and  Cole,  1936),  these 
hybrids  also  contain  a  hybrid  substance.  Because  of  the  cross  reactions  exist- 
ing between  these  two  hybrid  substances,  a  certain  degree  of  similarity  can 
be  assumed.  That  the  fraction  in  the  hybrid  substance  of  the  Fi-Pearlneck  X 

DIAGRAMMATIC  REPRESENTATION  OF  THE'HYBRID  SUBSTANCE" 
OF   THE  HYBRID  BETWEEN    PEARLNECK  AND  RINGDOVE 


VPEARLNECK>- 


^^■•^V-VV:- 


COMMON- >•:•.■•. '.<,"•.•••: 


RINGDOVE 


O  O  O   O   O 

0  0   O  O    O 

O  O  O    ^  Q 

o  O  O  OO 

O  0  0  0  o 


n    PEARLNECK 
^i   RINGDOVE 


HYBRID  SUBSTANCE 


(d-ll) 

Cd-i)                   (d^) 

M 

O  0«  OO 

O  O  o  o  o 
0  O  o  00 
O  0  £7  oO 
0  0  O   Oft 
o  a  o  OO 

0  0  a  00 
0  a  aao 

G    O    0  QtJ 
O    0    0  00 

dx-A 


d<-B 


dii-C 


MrBRIO  SUBSTANCE  ASSOCIATED 

WITH  SPECIFIC  PEARLNECK  SUBSTANCES 


RELATED   SUBSTANCES 


Fig.  15.2— The  separation  into  constituent  parts  of  the  hybrid  substance  of  the  species 

hybrid  between  Pearlneck  and  Ring  dove. 


Ring  dove,  which  is  primarily  if  not  entirely  responsible  for  the  cross  reac- 
tions, is  dx-A  may  be  deduced  from  Table  15.3,  in  that  this  fraction  (associ- 
ated with  d-ll)  is  the  only  one  which  will  exhaust  the  antibodies  from  the 
reagent  for  the  cells  of  the  pigeon-Ring  dove  hybrid  (column  5).  Also,  in 
unpublished  tests  the  reagent  for  the  hybrid  substance  of  the  pigeon-Ring 
dove  hybrid  (anti-hybrid  serum  absorbed  by  the  cells  of  pigeon  and  Ring 
dove)  did  not  react  with  Pearlneck  cells,  but  reacted  strongly  with  d-ll  cells, 
presumably  by  virtue  of  their  content  of  the  dx-A  fraction,  and  not  definitely 
with  cells  carrying  dx-B  or  dx-C.  If  the  dx-A  hybrid  substance  of  the  species 
hybrids  between  Pearlneck  and  Ring  dove  were  partially  or  largely  a  rear- 
rangement of  an  antigenic  substance,  in  this  case  d-ll,  which  is  species  spe- 
cific to  Pearlneck — since  the  Ring  dove  is  a  common  parent  of  the  two  kinds 


250  M.  R.  IRWIN 

of  species  hybrids — that  specific  substance  (cl-11)  should  be  detectable  in  the 
cells  of  livia. 

To  date,  reasonably  extensive  tests  (unpublished)  have  not  shown  that  the 
cells  of  livia  contain  more  than  a  trace  of  an  antigenic  substance  related  to  the 
d-11  of  Pearlneck.  Whatever  the  relationship  of  the  genes  in  Pearlneck  (as- 
sociated with  those  on  a  chromosome  effecting  the  d-11  specific  substance) 
and  livia,  respectively,  which  presumably  by  interaction  with  a  gene  or  genes 
from  Ring  dove  in  the  two  species  hybrids  effect  a  common  fraction  of  the 
two  hybrid  substances,  they  are  not  associated  with  genes  which  produce 
similar  antigenic  patterns  in  the  two  species.  On  these  grounds,  it  would  seem 
unlikely  that  the  hybrid  substances  in  these  two  kinds  of  species  hybrids  are 
merely  a  different  arrangement  of  a  species  specific  antigen. 

The  question  is  pertinent  as  to  whether  such  reactivities  in  the  cells  of 
these  other  species,  as  Senegal  and  Australian  crested,  are  themselves  an  in- 
dication of  antigenic  response  to  gene  interaction  within  each  species,  or  the 
more  direct  product  of  a  gene.  This  cannot  be  answered  directly.  But,  as 
given  in  Table  15.3,  the  fact  that  absorption  of  the  reagent  for  the  hybrid 
substance  by  fractions  dx-A,  dx-B,  or  dx-C  removes  the  antibodies  for  the 
cells  of  Senegal  indicates  that  there  is  some  common  constituent  of  these 
three  fractions  related  to,  if  not  identical  with,  a  reactive  substance  in  Senegal 
cells.  However,  only  the  dx-A  fraction  removes  the  antibodies  for  the  cells  of 
the  Australian  crested  dove.  Further,  absorption  by  the  cells  of  the  pigeon- 
Ring  dove  hybrid  also  removes  the  antibodies  from  this  reagent  for  the  cells 
of  the  Australian  crested  dove. 

The  hypothetical  explanations  could  be  advanced,  (1)  that  the  antigenic 
substances  in  Senegal  and  the  Australian  crested  dove,  themselves  being  dis- 
tinct, but  both  related  to  the  hybrid  substance  in  Pearlneck-Ring  dove 
hybrids,  are  the  result  of  a  genie  interaction.  But  there  is  no  evidence  for  such 
an  assumption.  Also,  (2)  the  argument  could  be  advanced  that  the  relation- 
ship between  these  substances  in  Senegal  and  Australian  crested,  and  in  the 
respective  species  hybrids,  is  fortuitous,  simulating  the  occurrence  of  the 
Forssman  antigen  in  many  species  of  animals  and  plants,  including  bacteria 
(Boyd,  1943).  That  is,  the  antigenic  substances  involved  (related  in  some 
manner  to  the  hybrid  substance)  may  be  gene  controlled  in  each  of  the  re- 
lated species,  since  indistinguishable  substances  to  those  of  Senegal  were 
found  in  four  other  species  of  Streptopelia,  capicola,  dussumierl,  humilis,  and 
orienlalis,  and  to  those  of  the  Australian  crested  dove  in  two  species  of  an- 
other genus,  Phaps  chalcoptera  and  Phaps  elegans,  but  the  antigenic  similarity 
to  the  hybrid  substance  is  by  virtue  of  some  related  antigenic  component. 
Various  ramifications  of  these  and  other  explanations  would  be  purely 
speculative. 

The  hybrid  substance,  as  it  has  been  observed  in  the  cells  of  various  species 
hybrids  in  birds,  simulates  for  cellular  antigens  the  expression  of  heterosis  in 
plants  and  animals.  That  is,  it  appears  as  the  resultant  of  an  interaction  be- 


SPECIFICITY  OF  GENE  EFFECTS 


251 


tween  genes.  One  may  well  ask  if  there  is  any  other  manifestation  of  heterosis 
in  these  species  hybrids  and  backcross  hybrids.  Extensive  measurements  of 
eight  body  characteristics,  as  over-all  length,  extent,  width  of  tarsus,  width  of 
band,  length  of  wing,  beak,  middle  toe,  and  tail,  were  made  over  a  7)eriod  of 
years  under  the  supervision  of  L.  J.  Cole.  The  differences  in  the  averages  of 
these  various  characteristics  between  the  parental  Pearlneck  and  Ring  dove 
species,  as  yet  unpublished,  were  statistically  significant,  and  the  averages  of 
the  measurements  of  these  characteristics  in  the  species  hybrids  showed  them 
to  be  in  general  intermediate  between  those  of  the  parental  species.  Thus 
there  was  no  evidence  of  heterosis  in  any  external  characteristic  of  the  species 
hybrids,  and  no  correlation  with  the  hybrid  substance  of  the  blood  cells. 

CELLULAR  CHARACTERS  WITHIN  A  SPECIES 

The  finding  that  one  or  more  genes  on  each  of  nine  or  ten  pairs  of  chromo- 
somes of  Pearlneck  had  effects  on  the  species  specific  antigens  of  the  blood 
cells  of  this  species  made  plausible  the  belief  that  many  more  genes  than 
commonly  believed  would  have  effects  within  a  species  making  for  individual- 
ity of  the  cellular  patterns.  Acting  on  this  assumption,  a  series  of  exploratory 
tests  w^ere  made  in  experimental  animals,  principally  in  cattle  and  chickens. 
For  example,  following  the  transfusion  of  the  blood  of  a  young  cow  into  her 
dam,  an  antibody  was  obtained  from  the  serum  of  the  recipient  which  re- 
acted (produced  lysis  of  the  reacting  cells  upon  the  addition  of  complement 
to  the  serum-cell  mixture)  with  the  cells  of  some  individuals,  but  not  with 
those  of  others.  The  reactive  substance  was  called  A. 

The  objective  was  to  be  able  to  detect  each  antigenic  factor  separately,  ac- 
cording to  the  following  criterion.  The  reactive  cells  from  any  individual 
should  remove  the  antibodies  from  the  reagent  specific  for  those  cells,  when 
added  in  excess  to  the  reagent.  However,  if  there  were  antibodies  in  the  re- 
agent which  recognized  two  or  more  distinct  blood  factors,  any  such  ab- 
sorption with  cells  containing  only  one  such  substance  would  remove  only  a 
part  of  the  antibodies.  Those  remaining  would  still  be  reactive  with  all  cells 
containing  the  substance  corresponding  to  the  unabsorbed  antibody. 

To  this  criterion  was  added  that  of  genetics  for  a  single  character,  using 
the  gene-frequency  method  since  controlled  matings  were  not  possible.  A 
typical  example  of  the  analysis  is  that  for  substance  A,  as  follows: 


Typk  ok 
Mating 

Number  of  offspring 

With 
Antigen  A 

Lacking 
Antigen  A 

AXA 

AX- 

217 

76 

0 

23 
51 

-X- 

41 

252 


M.  R.  IRWIN 


These  results  illustrate  the  observation  that  an  individual  has  any  cellular 
character  recognized  to  date  only  if  one  or  both  parents  possessed  it.  Also, 
each  behaved  as  if  it  were  a  dominant  to  its  absence. 

From  further  isoimmunizations  in  cattle,  and  from  immunizations  of  rab- 
bits, various  antisera  have  been  obtained  which  detect  other  antigenic  factors 
of  cattle  cells.  Each  of  these  has  been  subjected  to  the  criteria  of  both  genetics 
and  immunology  for  a  single  character,  as  described  in  reports  by  Ferguson 
(1941),  Ferguson  et  al.  (1942),  and  Stormont  (1950).  At  present,  about  forty 
different  reagents  are  regularly  used  in  typing  cattle  cells. 

Other   Antigens   in    Cattle 

As  stated  above,  the  first  substance  detected  in  cattle  cells  was  named  A. 
The  next  was  called  B,  the  next  C,  .  .  .  Z.  That  called  A'  implies  no  relation- 
ship to  A,  nor  B'  to  B,  etc.  Each  of  these  antigenic  factors  is  therefore  recog- 
nized independently,  and  when  subjected  to  an  analysis  of  gene  frequency, 
each  has  behaved  as  expected  if  effected  by  a  single  gene  in  comparison  to  its 
absence. 

However,  some  definite  associations  have  been  noted  among  them.  For 
example,  Ferguson  (1941)  reported  that  the  C  and  E  factors  were  not  inde- 
pendent, for  only  C  occurred  alone,  whereas  E  was  present  always  with  C, 
and  such  cells  therefore  had  CE.  It  was  postulated  that  there  were  three  al- 
lelic genes  involved,  one  for  the  components  C  and  E  together,  one  for  C 
alone,  and  a  third  for  the  absence  of  both  C  and  E. 

It  was  later  noted  by  Stormont  that  certain  additional  antigenic  factors 
appeared  only  if  one  or  more  other  components  also  were  present.  For  ex- 
ample, the  substance  B  occurs  alone,  as  does  that  called  G.  But  a  third  factor 
called  K  has  never  been  observed  unless  both  B  and  G  were  also  present. 
(A  possible  exception  to  this  rule  was  noted  shortly  after  these  factors  were 
first  demonstrable,  and  a  weak  reaction  at  that  test  with  the  reagent  for  the 
G  substance  was  probably  incorrectly  recorded.)  This  association  of  K  with 
B  and  G  has  been  noted  in  over  eighteen  hundred  animals  of  more  than  six 
thousand  tested.  Hence  the  combination  of  the  BGK  factors  has  always  oc- 
curred as  a  unit,  and  it  has  also  behaved  as  a  unit  in  the  progeny  of  individ- 
uals possessing  it.  A  compilation  of  some  unpublished  data  has  yielded  the 
following  information: 


Number  of  Offspring 

Type  of  Mating 

With 
BGK 

Without 
BGK 

BGKXBGK 

BGKX- 

151 

185 

0 

44 
137 

—  X  — 

160 

SPECIFICITY  OF  GENE   EFFECTS  253 

Notwithstanding  the  fact  tliat  H,  (1,  and  K  are  recognized  sei)arately  by- 
respective  reagents,  these  data,  and  the  observation  that  K  has  occurred 
only  with  both  H  and  G,  are  strong  evidence  for  tlie  conclusion  that  B,  G,  and 
K  in  the  cells  behave  as  a  unit. 

Further,  offsj)ring  of  some  individuals  ])ossessing  B  and  G  (BG)  in  their 
cells  have  given  only  two  classes  of  offspring,  those  with  B  and  those  with  G, 
as  would  be  expected  if  the  causative  genes  were  alleles.  But  another  type  of 
BG  individual  has  produced  offspring  of  two  quite  different  types — those 
with  both  B  and  G  (BG)  and  those  with  neither,  as  if  a  gene  producing  B  and 

TABLE  15.4 

THE  DISTRIBUTION  OF  THE  CONSTITUENTS  OF 

THE  "B"  COMPLEXES  IN  THE  OFFSPRING 

OF  SELECTED  SIRES 


Sire 

Antigenic 
Complex 

Number 
of  Off- 
spring 

Antigenic 
Complex 

Number 
of  Off- 
spring 

H-1 

H-4 

H-5 

H-6 

H-7 

H-11 

H-19 

G-19 

BbgioiTza' 

Bb02A'E3 
BbOiY2D' 

BbOi 
Bbgke'. 
Boia' 
Bgy2e; 

BlE'i 

25 
35 
26 
15 
14 
31 
19 
8 

BoiYjA' 
Bo3J'k' 
Bo3j'k' 

BgYjE'i 

BgyoeJ 

B03I'k' 

Bb 
Bb 

23 
31 
24 

23 
15 
23 
13 

7 

G  together  was  allelic  to  one  not  effecting  either  B  or  G.  These  combinations 
of  antigenic  substances,  as  BG  and  BGK,  have  been  called  antigenic  com- 
plexes. 

There  are  two  series  of  such  complexes,  called  the  B  and  C  series,  respec- 
tively. In  the  B  series  there  are  twenty-one  of  the  forty-odd  antigenic  char- 
acters which  are  associated  in  various  conbinations.  At  least  seven  of  these 
may  appear  singly,  as  was  described  for  B  and  G.  The  other  fourteen  have 
been  found  only  in  various  antigenic  complexes,  each  of  which  may  be  made 
up  of  from  two  to  eight  of  the  twenty-one  characters.  The  majority  of  these 
twenty-one  characters  do  not  occur  at  random  in  a  complex  with  each  of  the 
others.  As  was  stated  above,  the  character  K  has  always  been  found  with  B 
and  G,  but  it  has  never  occurred  with  I,  with  which  it  appears  as  a  contrast- 
ing substance.  In  contrast,  either  B  or  G  may  be  present  in  a  complex  with  I. 
No  separation  of  the  antigenic  characters  of  a  complex  has  ever  been  ob- 
served in  the  cells  of  the  offspring  of  an  individual  possessing  it.  A  few  ex- 
amples are  listed  in  Table  15.4  from  more  complete  data  given  in  a  paper  by 
Stormont,  Owen,  and  Irwin  (1951).  All  present  evidence  makes  it  seem  some- 
what more  reasonable  to  assume  that  each  antigenic  complex  is  produced  by 
a  single  gene  than  by  linked  genes.  The  various  antigenic  complexes  in  each 


254  M.  R.  IRWIN 

of  the  two  systems,  or  series,  would  then  be  produced  by  a  series  of  multiple 
alleles.  The  possibility  of  pseudo-alleles  cannot  be  eliminated,  but  for  the 
present  may  be  assumed  not  to  be  a  complicating  factor. 

If  the  assumption  be  granted  that  a  single  gene  controls  an  antigenic  com- 
plex, as  BGK,  what  explanation  or  explanations  can  be  proposed  for  the  dif- 
ferent antigenic  specificities  of  this  and  other  complexes,  and,  in  turn,  what 
can  be  inferred  from  such  an  explanation  as  to  the  action  of  the  causative 
gene? 

Antigens    of    Pneumococci 

By  virtue  of  the  ability  to  attach  simple  chemical  compounds  to  proteins, 
thereby  preparing  conjugated  antigens  with  specifically  reacting  components 
of  known  constitution,  there  has  emerged  from  such  studies  the  realization 
that  a  so-called  single  antigenic  substance  may  engender  a  multiplicity  of 
antibodies  of  varying  specificities  (see  Landsteiner,  1945,  for  a  critical  review 
and  references).  A  pertinent  example  of  this  sort  may  be  found  in  the  anti- 
genic relationship  existing  between  type  III  and  type  VIII  pneumococci. 
Cross  reactions  between  the  respective  antisera  (produced  in  horses)  and  the 
two  types  of  pneumococci  have  been  observed,  implying  to  them  some  sort  of 
antigenic  similarity. 

As  is  well  known,  the  specificities  of  the  pneumococcal  types  depend  upon 
the  carbohydrates  of  the  capsules  (Heidelberger  and  Avery,  1923,  1924). 
Thus,  the  carbohydrate  of  type  III  has  been  found  to  be  a  polyaldobionic 
acid  (Reeves  and  Goebel,  1941).  The  understanding  of  the  structure  of  the 
polysaccharide  of  type  VIII  is  not  as  complete  as  for  type  III,  but  about  60 
per  cent  of  the  molecule  of  the  carbohydrate  of  type  VIII  appears  to  be 
aldobionic  acid.  Cross  reactivity  may  therefore  be  expected  between  the 
soluble  specific  substances  of  types  III  (S  III)  and  VIII  (S  VIII),  by  virtue 
of  the  presence  in  each  of  multiples  of  the  same  aldobionic  acid  as  a  structural 
unit.  It  is  probable  that  the  serologically  reactive  unit  in  each  of  these  two 
types  is  a  larger  portion  of  the  polysaccharide  molecule  than  a  single  chemical 
structural  unit.  Type  S  VIII  also  contains  approximately  two  glucose  mole- 
cules for  every  aldobionic  acid  residue,  thereby  presumably  accounting  for  at 
least  a  part  of  the  specificity  of  type  VIII  in  contrast  to  type  III.  Thus  it  may 
be  seen  that  serological  cross  reactions  may  be  expected  when  the  antigenic 
substances  under  comparison  are  closely  related  chemically.  Also  to  be  ex- 
pected is  the  ability  to  distinguish  between  such  substances,  as  was  actually 
possible  in  the  case  of  types  III  and  VIII  (Heidelberger,  Kabat,  and  Meyer, 

1942). 

Genetic  Significance 

The  above  example  may  be  combined  with  other  findings  in  the  field  of  im- 
munochemistry  to  allow  the  statement  that  antigenic  substances  of  related 
but  not  identical  chemical  constitution  may — but  sometimes  do  not — incite 
the  production  of  cross  reacting  antibodies.  From  the  serological  point  of 


SPECIFICITY  OF  GENE  EFFECTS  255 

view,  a  pertinent  question  concerning  tiiese  antigenic  complexes  in  cattle  is 
whether  the  cells  which  react  with  the  B  reagent,  or  with  any  other  specific 
reagent,  do  so  by  virtue  of  the  presence  of  a  specific  reacting  substance  in  a 
single  antigenic  molecule,  or  otherwise?  Does  the  comj)lex  BOX,  for  exam])le, 
represent  (1)  three  different  and  separate  antigenic  substances?  Or  does  it 
represent  (2)  a  single  antigenic  substance  with  (a)  a  possible  common  base 
and  three  more  or  less  different  reactive  groups  accounting  for  B,  G,  and  K, 
respectively,  or  (6)  a  single  substance  capable  of  inciting  many  specificities  of 
antibodies,  of  which  those  for  B,  G,  and  K  represent  only  a  part  of  the  re- 
activities of  the  spectrum  of  antibodies  which  may  be  produced?  A  combina- 
tion of  (a)  and  (b)  also  may  be  a  possibility. 

At  present,  very  little  experimental  evidence  is  available  concerning  the 
adequacy  of  any  one  or  combination  of  the  above  possibilities  to  explain  the 
antigenic  relationships  of  the  components  of  the  antigenic  complexes  of  cattle 
cells.  Tests  are  under  way  to  determine  whether  the  reactive  substance  called 
B,  for  example,  is  the  same  in  all  cells  in  which  it  appears,  whether  singly  or 
in  an  antigenic  complex. 

In  terms  of  the  action  of  the  causative  genes,  apart  from  the  possibilities  of 
linkage  and  pseudo-allelism,  the  question  seems  to  resolve  itself  around  two 
main  aspects:  (1)  Do  the  genes  controlling  an  antigenic  complex,  as  a  single 
gene  for  BGK,  have  separate  specificities  for  B,  G,  and  K,  or  (2)  does  this 
gene  produce  a  single  substance  with  no  such  separate  specificities,  and  the 
similarities  between  such  a  complex  as  BGK  and  BGIY,  are  due  primarily  if 
not  entirely  to  the  general  similarities  in  their  chemical  structure.  The  writer 
is  inclined  to  adopt  a  combination  of  these  two  possibilities  as  a  current  work- 
ing hypothesis.  No  matter  what  may  eventually  prove  to  be  the  correct  in- 
terpretation of  antigenic  structure  of  the  complexes,  and  the  action  of  the 
controlling  genes,  it  appears  that  these  studies  have  given  some  insight  into 
the  complexities  of  the  gene  products  and  perhaps  also  of  the  causative  genes. 

The  studies  of  the  specificities  of  the  gene  products — the  antigens  of  the 
blood  cells  of  cattle — and  the  resulting  inferences  of  the  structure  of  the  genes 
themselves,  may  not  be  directly  related  to  the  over-all  heterosis  problem. 
Nevertheless  the  writer  is  convinced  that  somewhat  comparable  specificities 
might  well  be  obtained  in  plants,  in  which  attempts  are  currently  in  progress 
to  measure  various  aspects  of  the  genetic  bases  for  heterosis.  Just  how  useful 
an  additional  tool  of  this  sort  would  he  is  only  a  guess. 


CARL  C.  LINDEGREN 

Southern  Illinois  Universify 


Chapter  16 

Genetics  and  Cytology 
of  Soccharomyces 


In  the  middle  of  the  last  century,  Buchner  ground  up  yeast  cells  and  proved 
that  the  cell-free  filtrate  contained  a  substance  capable  of  fermenting  sugar. 
This  experiment  settled  a  heated  controversy  between  Liebig  and  Pasteur 
concerning  whether  or  not  living  structures  were  essential  to  fermentation. 
The  substance  responsible  for  the  fermentation  was  called  an  enzyme,  the 
word  being  derived  from  the  Greek  and  meaning  "in  yeast."  Since  that  time, 
yeast  has  been  the  organism  of  choice  for  experimenting  in  enzyme  chemistry 
because  of  the  abundant  supply  obtainable  from  breweries  and  from  factories 
producing  bakers'  yeast.  The  biochemistr}^  of  fermentation  has  provided  the 
foundation  for  our  present  understanding  of  the  biochemistry  of  respiration 
and  of  muscular  contraction — two  of  the  fundamental  problems  which  have 
intrigued  biologists.  It  has  led  to  an  understanding  of  vitamins  and  through 
them  to  an  understanding  of  chemotherapy. 

BIOCHEMICAL  DEFECTS  AS  GENE  MARKERS 
The  work  of  Beadle  and  Tatum  has  popularized  the  generally  accepted 
view  that  enzymes  are  derived  somehow  or  other  from  genes.  Their  work 
initiated  a  new  interest  in  biochemical  genetics.  They  showed  that  the  in- 
activation  of  a  specific  gene  caused  a  deficiency  which  could  be  met  by  sup- 
plying a  specific  chemical.  Vitamins,  amino  acids,  purines,  and  pyrimidines 
were  the  substances  chosen  in  this  analysis.  They  used  the  fungus,  Neuro- 
spora,  because  its  life  cycle  had  been  thoroughly  worked  out  by  B.  O.  Dodge 
and  because  the  Lindegrens  had  shown  by  genetical  analysis  that  it  contained 
conventional  chromosomes  on  which  genes,  arranged  in  linear  order,  could 
be  mapped  by  the  standard  procedures  used  in  studying  corn  and  the  fruit 

fly- 

256 


GENETICS  AND  CYTOLOGY  OF  SACCHAROMYCES  257 

YEAST  GENETICS 

Until  1935,  yeasts  were  considered  to  be  devoid  of  sex  and,  therefore,  un- 
suitable for  genetical  analysis.  At  that  time,  Winge  showed  that  the  standard 
yeast  cell  carried  two  sets  of  chromosomes — one  contributed  from  each 
parent — and  was,  therefore,  a  typical  hybrid.  The  hybrid  yeast  cell  produces 
four  spores,  each  with  a  single  set  of  chromosomes.  Each  of  these  spores  is  a 
sex  cell.  By  fusing  in  pairs  they  can  produce  the  standard  (hybrid)  yeast  cell 
and  complete  the  life  cycle.  In  this  laboratory  it  was  shown  that  the  spores 
are  of  two  mating  types,  and  that  each  spore  can  produce  a  culture  each  cell 
of  which  can  act  as  a  sex  cell,  like  the  original  spore.  Mass  matings  between 
two  such  spore-cultures  result  in  the  production  of  fusion  cells,  from  which 
new  hybrids  are  produced  by  budding. 

This  work  made  it  possible  to  study  the  inheritance  of  biochemical  de- 
ficiencies in  the  organism  on  which  classical  enzyme  study  is  based,  and  to 
attack  the  problem  of  the  relation  of  genes  to  enzymes  in  this  fruitful  mate- 
rial. We  have  related  specific  genes  to  several  of  the  most  thoroughly  studied 
classical  enzymes:  sucrase,  maltase,  alpha  methyl  glucosidase,  melibiase,  and 
galactase. 

The  principal  advantages  of  yeasts  for  biochemical  genetics  are: 

(1)  Yeast  enzymes  have  been  the  subject  of  intensive  biochemical  study. 

(2)  Techniques  for  studying  respiration  and  fermentation  are  based  prin- 
cipally on  work  with  yeast  and  thus  especially  adapted  to  this  organism. 
Yeasts  grow  as  free  cells  rather  than  as  mycelial  matts  and,  therefore,  can  be 
subdivided  any  number  of  times  without  injury,  thus  simplifying  weighing 
and  dilution  of  the  cells. 

(3)  Large  quantities  of  cells  are  available  from  industrial  sources  or  can  be 
grown  cheaply  and  quickly  and  are  easily  stored  in  living  condition. 

(4)  A  variety  of  genes  concerned  with  the  differential  utilization  of  nu- 
merous monoses  as  well  as  di-  and  poly-saccharides  are  available. 

(5)  A  polyploid  series  of  yeast  cultures  is  now  available:  (a)  haploid  cells, 
each  containing  a  single  set  of  chromosomes,  (b)  diploid  yeast  cells,  each  con- 
taining the  double  number  of  chromosomes,  (c)  triploid,  and  (d)  tetraploid 
cells  (made  available  by  our  recent  discovery  of  diploid  gametes  [Lindegren 
and  Lindegren,  1951]). 

(6)  With  free  cells  it  is  possible  to  study  competition  between  genotypes 
and  to  observe  the  advantages  or  disadvantages  in  controlled  environments. 
The  populations  involved  are  enormous  and  the  life  cycles  short,  so  it  is  pos- 
sible to  simulate  natural  selection  in  the  laboratory.  Experiments  of  this  type 
have  enjoyed  an  enormous  vogue  with  bacteria,  but  it  has  not  been  possible 
to  distinguish  gene-controlled  variation  from  differentiation.  Eor  this  reason, 
experiments  with  bacteria  cannot  be  interpreted  in  terms  of  the  comparison 
between  gene-controlled  and  other  types  of  inherited  characteristics. 


258 


CARL  C.  LINDEGREN 


CHROMOSOMAL  INHERITANCE 

In  our  selected  breeding  stocks  of  Saccharomyces,  irregular  segregations 
do  not  occur  very  frequently.  In  maize  or  Drosophila  a  similar  frequency  of 
irregularity  would  not  be  detectable  since  tetrad  analysis  is  not  possible  in 
these  forms.  Using  regularly  segregating  stocks  of  Saccharomyces  we  have 
mapped  four  and  possibly  five  chromosomes  for  genes  controlling  the  fer- 
mentation of  carbohydrates  and  the  synthesis  of  various  nutrilites  (Fig. 
16.1). 


HI 


AN 


PN 


AD  I 


IN 


PY 


TH 


24 


23 


22 


10 


30 


24 
AD  2 


26 


ME 

l_ 


40 


45 


I 


22 


PB 


22 


UR 


Fig.  16.1     Chromosome  maps  of  Saccharomyces. 

Chromosome  I,  PN  (pantothenate),  centromere,  ADl  (adenine),  IN 
(inositol),  PY  (pyridoxine),  and  TH  (thiamin). 

Chromosome  II,  centromere,  G  (galactose),  AD2  (adenine),  ME  (meli- 
biose). 

Chromosome  III,  centromere,  a  (mating  type). 

Chromosome  IV,  centromere,  PB  (paraminobenzoic  acid). 

Chromosome  V,  centromere,  UR  (uracil). 

Chromosomes  IV  and  V  may  or  may  not  be  different;  UR  and  PB  have 
not  been  used  in  the  same  hybrid. 

HI  (histidine)  and  AN  (anthranilic  acid)  are  linked  to  each  other  (24 
morgans)  but  have  not  yet  been  located  on  a  chromosome. 

DIRECT  TETRAD  ANALYSIS 

The  focal  point  in  the  life  cycle  is  the  reduction  division,  at  which  the 
chromosomes  of  a  diploid  cell  are  sorted  out,  and  the  haploid  sex  cells  (such  as 
sperm,  eggs,  pollen,  or  yeast  spores)  are  produced.  Each  diploid  parent  cell 
divides  twice  to  produce  a  tetrad  of  four  haploid  sexual  nuclei.  This  process 
is  substantially  the  same  whether  a  single  yeast  cell  produces  four  spores  or  a 
cell  in  the  testis  produces  four  sperm.  In  yeast,  however,  each  of  the  four 
spores  of  a  single  tetrad  can  produce  clones  which  are  available  for  individual 
study,  and  the  reduction  division  can  be  analyzed  directly  instead  of  by  in- 
ference. 


GENETICS  AND  CYTOLOGY  OF  SACCHAROMYCES  259 

Many  yeast  hybrids  have  been  produced  by  mating  sex  cells  carrying 
chromosomes  marked  with  biochemical  mutant  genes.  The  tetrads  from  these 
hybrids  have  been  analyzed  by  growing  clones  from  each  of  the  four  spores 
of  a  single  ascus  and  classifying  each  of  the  spore-cultures.  These  exj)eriments 
are  direcl  tests  of  the  Memieliau  theory.  They  have  shown  that  excei)tions  to  the 
Mendelian  theory  occur  more  frequently  than  was  hitherto  supposed. 

CURRENT  STATUS  OF  IRREGULAR  MENDELIAN  SEGREGATION 

Tetrad  analysis  of  triploid  and  tetraploid  yeasts  has  revealed  that  some  of 
the  irregular  (not  2 : 2)  segregations  in  hybrid  asci  arise  from  the  fact  that  one 
or  both  of  the  parents  is  diploid  (Lindegren  and  Lindegren,  1951;  Roman, 
Hawthorne,  and  Douglas,  1951).  Roman,  Hawthorne,  and  Douglas  have 
concluded  that  all  irregular  segregations  in  Saccharomyces  arise  from  the 
segregation  of  triploid  or  tetraploid  zygotes.  We  have  recently  completed  the 
analysis  of  segregation  in  diploid  hybrids  heterozygous  for  both  MA /ma  and 
MG/mg.  This  analysis  revealed  that  in  many  asci  in  which  segregation  of 
MA /ma  was  2:2  {MA  MA  ma  ma),  segregation  of  MG/mg  was  1:3  {MG  mg 
mg  mg).  This  finding  excludes  the  possibility  that  the  hybrid  was  either 
triploid  or  tetraploid  since  segregation  of  both  genes  would  have  been  equally 
affected.  The  phenomenon  has  been  explained  as  conversion  of  the  MG  gene 
to  mg  in  the  zygote.  This  conclusion  is  further  supported  by  evidence  indicat- 
ing that  both  genes  are  in  the  same  linkage  group. 

One  hypothesis  of  the  nature  of  the  gene  developed  during  the  study  of 
irregular  segregation  seems  to  have  some  merit.  This  is  the  proposal  that  the 
gene  is  a  complex  of  many  more  or  less  loosely  connected  molecules  rather 
than  a  single  macromolecule.  In  this  view,  the  gene  is  composed  of  a  series  of 
identical  sites  around  the  periphery  of  a  more  or  less  cylindrical  chromosome. 
These  sites  may  be  extremely  numerous  since  they  are  of  molecular  dimen- 
sions around  the  periphery  of  a  thread  easily  visible  under  the  microscope. 
At  these  sites  identical  agents  responsible  for  the  action  of  the  gene  are 
located. 

GENE  DIVISION 

The  concept  of  the  gene  as  a  bracelet  of  catalysts  arranged  on  the  outside 
of  the  chromosome  simplifies  the  concept  of  gene  reproduction.  When  one 
conceived  of  genes  as  macromolecules  arranged  like  beads  on  a  string,  it  was 
difficult  to  understand  how  all  the  genes  on  a  chromosome  could  divide 
simultaneously.  If,  however,  there  are  thousands  of  loci  and  chromogenes  at 
the  site  of  a  single  gene  on  the  outside  of  an  otherwise  inert  chromosome  which 
is  composed  principally  of  skeletal  material,  any  longitudinal  splitting  of  the 
chromosome  will  partition  two  qualitatively  equivalent  parts  which  may  or 
may  not  be  quantitatively  equivalent.  The  restoration  of  balance  by  inter- 
dependence of  the  autonomous  organelles  may  make  precise  division  unneces- 
sary. 


260 


CARL  C.  LINDEGREN 


EXTRACHROMOSOMAL  INHERITANCE 

When  a  pure  haplophase  culture  of  red  yeast  (adenine  dependent)  is 
planted  on  an  agar  plate,  both  red  and  white  colonies  appear.  When  the  white 
colonies  are  subcultured,  only  white  colonies  appear.  The  red  cells  when 
planted  on  a  second  plate  continue  to  produce  both  red  and  white  colonies. 
The  white  colonies  are  stable  variants  derived  from  red.  Bacterial  variations 
of  this  type  are  ordinarily  called  gene  mutations,  but  bacteriologists  have 
been  unable  to  test  their  so-called  mutations  by  breeding  experiments  except 


While  Normal 
(Adenine  Independent) 


Red  Original  Culture 
(Adenine  Dependent) 


White  Normal 
(Adenine  Independent) 


Red  Subculture 
(Adenine  Dependent) 


White  Variant 
(Adenine  Dependent) 


Red  Subculture  White  Variant  White  Subculture 

Normal  X (Adenine  Dependent)  (Adenine  Dependent)  (Adenine  Dependent)  X  Normal 


Hybrid 

(White  Culture,  Adenine  Independent) 

ASCI  Analyzed 
White  (Adenine  Independent) 
White  (Adenine  Independent) 
Red  (Adenine  Dependent) 
Red  (Adenine  Dependent) 


Hybrid 

(White  Culture,  Adenine  Independent) 

ASCI  Analyzed 
White  (Adenine  Independent) 
White  (Adenine  Independent) 
Red  (Adenine  Dependent) 
Red  (Adenine  Dependent) 


Fig.  16.2 — Inheritance  of  pink  versus  white  colony  in  Saccharomyces.  The  white  mutants 
derived  from  pink  produce  pink  offspring  and  are  indistinguishable  from  the  original  pink 

genetically. 

in  a  few  cases.  The  white  cultures  have  lost  their  color  but  they  are  still 
characteristically  adenine  dependent  like  their  red  progenitor.  Breeding  ex- 
periments (Fig.  16.2)  have  been  carried  out  with  the  white  yeast  cultures 
derived  from  the  red.  When  the  derived  white  cultures  were  used  as  parents, 
they  produced  precisely  the  same  kind  of  offspring  as  the  original  red  culture 
from  which  they  arose.  This  proves  that  the  change  from  red  to  white  did 
not  affect  a  gene.  The  change  from  red  to  white  may,  therefore,  be  called  a 
differentiation  since  it  occurs  without  gene  change. 

The  phenomenon  of  Dauermodifikation  which  was  first  described  by  JoUos 
(1934)  has  thus  been  confirmed  in  yeast  genetics.  The  stable  change  from 


GENETICS  AND  CYTOLOGY  OF  SACCHAROMYCES  261 

red  to  white  resembles  the  discontinuous  variations  which  occur  in  the  vege- 
tative cycles  of  bacteria.  Hybridization  experiments  have  revealed  that  the 
origin  of  white  cultures  does  not  involve  gene  change.  This  phenomenon  in 
yeast,  called  depletion  mutation,  is  identical  with  Dauermodifikation  in  Para- 
mecia.  Since  neither  involves  gene  change,  both  are  equivalent  to  differentia- 
tion. 

It  is  not  possible  to  study  Dauermodifikationen  using  the  classical  objects 
of  genetical  research,  maize  and  Drosophila,  since  each  generation  of  these 
higher  organisms  is  produced  sexually — a  process  during  which  Dauermodi- 
fikationen revert  to  normal.  The  stable  variants  in  vegetative  cultures  of 
yeast,  which  revert  to  normal  (produce  only  normal  offspring  by  sexual  re- 
production) have  no  parallel  in  maize  and  Drosophila.  This  points  up  a 
striking  disadvantage  of  maize  and  Drosophila — that  they  cannot  be  propa- 
gated vegetatively.  One  cannot  be  certain  that  the  characteristic  variations 
in  flies,  which  occur  when  they  hatch  on  wet  medium  or  are  subjected  to 
shock  treatment,  would  be  lost  on  vegetative  cultures  unless  one  were  able 
to  propagate  the  bent  wings  or  other  peculiarities  asexually,  possibly  in  tis- 
sue culture. 

THE  AUTONOMOUS  ORGANELLES  OF  THE  YEAST  CELL 

In  addition  to  the  chromosomes  (Lindegren,  1949)  there  are  other  perma- 
nent structures  in  the  yeast  cell  which  never  originate  de  novo  (Lindegren, 
1951).  They  have  the  same  type  of  continuity  in  time  as  chromosomes  but 
are  less  precisely  partitioned  than  the  latter. 

The  Cytoplasm 

The  cytoplasm  is  a  limpid  fluid  which  is  transmitted  to  each  daughter  cell. 
It  is  rich  in  RNA  but  varies  in  basophily  and  contains  the  mitochondria, 
usually  adhering  to  the  surface  of  the  centrosome  or  the  nuclear  vacuole. 

The  Mitochondria 
The  state  of  the  mitochondria  varies  from  highly  refractile  lipoidal  struc- 
tures, sharply  defined  from  the  cytoplasm  to  less  refractile  organelles  with 
somewhat  irregular  boundaries. 

The  Centrosome 

The  centrosome  is  a  solid  and  rigid  structure  which  stains  with  acid 
fuchsin  but  does  not  stain  with  basic  dyes.  This  highly  basic  organelle  may 
contain  some  of  the  basic  proteins  which  Caspersson  and  Mirsky  have  found 
in  chromosomes.  The  centrosome  is  always  attached  to  the  nuclear  vacuole 
and  is  the  most  rigid  structure  in  the  cell  as  revealed  by  its  behavior  following 
shrinkage  of  the  cell.  It  never  originates  de  novo  and  plays  a  leading  part  in 
budding,  copulation,  and  meiosis. 


262  CARL  C.  UNDEGREN 

The  Centrochromatin 

The  centrochromatin  is  a  basophilic,  Feulgen-positive  substance  closely 
attached  to  the  basic  centrosome  (probably  by  an  acid-base  reaction).  Some 
portion  of  it  is  usually  in  contact  with  the  nuclear  vacuole.  It  is  partitioned 
between  the  cells  following  budding  by  a  direct  division  controlled  by  two 
tiny  centrioles.  In  the  resting  cell  it  may  assume  a  spherical  form  and  cover 
most  of  the  centrosome.  In  division  it  is  usually  present  in  the  form  of  a  long 
strand.  The  centrosome  and  centrochromatin  have  been  identified  with  the 
nucleus  by  several  workers,  but  this  view  has  been  criticized  by  Lindegren 
(1949),  Lindegren  and  Rafalko  (1950),  and  Rafalko  and  Lindegren  (1951). 
The  filament  often  bends  on  itself  to  assume  a  V-  or  U-shape.  In  some 
preparations  it  appears  to  be  composed  of  numerous  small  particles,  but  this 
is  due  to  poor  fixation  and  is  especially  prevalent  in  preparations  fixed  with 
alkali.  The  view  that  the  centrochromatin  is  a  single  filament  external  to 
the  centrosome  is  supported  by  a  multitude  of  observations  on  well-fixed 
cells.  Centrochromatin  is  probably  homologous  to  the  heterochromatin  of 
higher  forms  differing  only  in  being  carried  on  the  centrosome  rather  than 
the  chromosome. 

The  Nuclear  Membrane  and  the  Chromosomes 

The  nuclear  vacuole  contains  the  chromosomes  and  the  nucleolus.  The 
chromosomes  are  partitioned  between  mother  and  bud  vacuole  in  a  precise 
orderly  manner  without  recourse  to  a  spindle.  The  wall  of  the  nuclear  vacuole 
does  not  break  down  at  any  time  in  the  life  cycle;  it  is  a  permanent  cellular 
structure. 

The  Cell  Membrane  and  the  Cell  Wall 

The  cell  membrane  is  a  permanent  cell  structure.  The  cell  wall  appears  to 
be  formed  de  novo  in  the  spores,  but  it  may  depend  on  the  cell  wall  sur- 
rounding the  ascus  for  its  origin. 

BUDDING 

Figure  16.3-1  shows  a  cell  in  which  the  acidophilic  centrosome  attached 
to  the  nuclear  vacuole  is  surrounded  by  the  darkly  staining  cytoplasm.  A 
band  of  basophilic  centrochromatin  is  securely  applied  to  the  side  of  the 
centrosome  and  is  also  in  contact  with  the  nuclear  vacuole.  Greater  differen- 
tiation often  reveals  a  small  centriole  at  each  end  of  this  band.  The  nuclear 
contents  are  unstained. 

Figure  16.3-2  shows  the  first  step  in  the  process  of  budding.  The  centro- 
some produces  a  small  conical  process  which  forces  its  way  through  the 
cytoplasm  and  erupts  into  the  new  bud  shown  in  Figure  16.3-3. 

Figure  16.3-4.  The  nuclear  vacuole  sends  out  a  long,  slender  process  which 
follows  the  centrosome  into  the  bud.  Although  the  cell  wall  is  not  visible  in 
these  preparations  it  must  be  assumed  that  the  cell  wall  never  ruptures  but  is 


.Mk 


i 


7 


8 


10 


12  . 


Fig.  16.3 — Behavior  of  the  centrosome  and  centrochromalin  during  budding. 


264  CARL  C.  LINDEGREN 

extended  to  enclose  the  bud  at  all  times.  The  vacuolar  process  follows  the 
external  surface  of  the  centrosome  into  the  bud,  lying  between  the  cell  wall 
and  the  centrosome. 

Figure  16.3-5  shows  a  cell  in  which  the  bud  vacuole  has  received  its  two- 
stranded  chromosome  complex.  This  is  an  exception  to  the  rule  that  the 
chromosomes  usually  are  completely  destained  in  the  differentiation  by 
iron  alum. 

In  Figure  16.3-6  the  bud  vacuole  is  lobed.  This  is  a  rather  common  phe- 
nomenon. The  cytoplasm  has  passed  into  the  bud  and  completely  surrounds 
the  centrosome  and  the  bud  vacuole.  The  extension  of  the  centrochromatin 
along  the  surface  of  the  acidophilic  centrosome  has  begun. 

Figures  16.3-7,  16.3-8,  and  16.3-9  show  cells  in  which  the  separation  of 
the  centrochromatin  has  been  completed  with  mother  and  bud  held  to- 
gether by  the  centrosome. 

In  Figure  16.3-10  the  division  of  the  centrosome  is  complete,  but  both 
centrosomes  are  near  the  point  of  budding.  In  Figure  16.3-11  the  bud  cen- 
trosome has  reached  the  distal  end  of  the  bud  while  the  mother  cell  centro- 
some still  lies  in  the  neighborhood  of  the  point  of  budding.  In  Figure  16.3-12 
both  centrosomes  have  reached  the  distal  ends  of  the  cells  and  are  prepared 
for  the  formation  of  the  next  bud. 

CONCLUSIONS  CONCERNING  EXTRACHROMOSOAAAL  INHERITANCE 

Cytological  examination  of  the  yeast  cell  shows  that  many  of  its  organelles 
may  have  the  same  integrity  and  continuity  in  time  that  characterize  the 
chromosomes — they  cannot  arise  de  novo.  In  the  yeast  cell  there  are  seven 
or  eight  such  "continuous"  organelles.  The  cell  membrane,  the  nuclear 
membrane,  the  centrosome,  the  centrochromatin,  the  cytoplasm,  the  mito- 
chondria, and  the  chromosomes  are  permanent  cell  structures.  Because  they 
apparently  divide  in  a  manner  which  does  not  provide  for  precise  trans- 
mission of  specific  portions  to  each  daughter  cell,  it  appears  that  the  other 
components  differ  from  the  chromosomes  in  a  significant  manner — they  are 
probably  homogeneous,  or  their  heterogeneity  is  simple,  possibly  a  few 
different  types  of  dipolar  molecules  held  together  in  a  specific  manner. 

There  is  no  reason  to  assume  that  any  one  of  these  components  is  of  more 
importance,  or  directs  the  "activities"  of  any  one  of  the  other  components. 
The  cell  can  function  only  if  all  its  component  parts  are  present  in  proper 
structural  correlation  and  in  adequate  amounts.  There  is  no  reason  to  as- 
sume that  any  one  of  these  components  is  unique  in  the  manner  in  which  it 
reproduces  itself.  The  present  hypothesis  proposes  that  they  all  reproduce 
by  the  simple  accretion  of  molecules  like  those  which  they  contain,  and  it  is 
their  association  with  each  other  in  an  adequate  milieu  which  provides  the 
molecules  necessary  for  their  increase  in  size.  Each  of  the  different  organelles 
is  rate  limiting  in  growth.  When  any  one  is  present  in  less  than  the  minimal 


GENETICS  AND  CYTOLOGY  OF  SACCHAROMYCES  265 

amount,  the  other  organelles  cannot  obtain  the  supply  of  molecules  necessary 
for  maintenance  and  increase  until  the  amount  of  the  deficient  organelle  has 
increased. 

The  chromosomes  differ  from  the  other  permanent  organelles  in  their  high 
degree  of  linear  heterogeneity.  It  is  this  characteristic  which  has  given  them 
the  spurious  appearance  of  "controlling"  other  cellular  activities.  Mutations 
with  which  we  are  familiar  in  the  laboratory  constitute  defects  or  deletions 
in  the  extraordinarily  heterogeneous  chromosomes.  The  deficiency  in  the 
organism  caused  by  the  defect — the  deletion  of  the  contribution  ordinarily 
made  by  the  intact  region  of  the  chromosome — becomes  apparent  only  be- 
cause the  rest  of  the  chromosome  produces  sufficient  materials  to  enable  the 
defective  cell  to  continue  to  grow  in  its  absence,  although  in  a  manner  differ- 
ent from  that  which  was  previously  characteristic. 

Any  transmissible  defect  in  a  homogeneous  structure  like  the  cell  wall, 
the  cell  membrane,  the  nuclear  membrane,  the  centrosome,  or  the  centro- 
chromatin  would  result  in  total  failure  of  the  organism  to  survive  and  bring 
all  vital  activity  to  a  halt.  The  survival  of  the  defective  mutants  in  their 
altered  condition  due  to  the  defect  in  the  chromosome  (which  has  been  called 
a  mutant  gene)  has  led  to  the  view  that  genes  are  different  from  other  cellu- 
lar components  since  they  can  reproduce  variations  in  themselves.  This  is  an 
incorrect  point  of  view.  It  is  more  proper  to  say  that  when  a  defect  or  dele- 
tion occurs  in  a  small  segment  of  a  chromosome,  the  rest  of  the  organism  can 
carry  on,  albeit  in  a  changed  condition  due  to  the  absence  of  the  contribution 
previously  made  by  that  region,  now  called  the  gene.  This  denies  the  im- 
portance of  the  ordinary  mutations  encountered  in  the  laboratory  as  factors 
for  progressive  evolution,  and  implies  that  progress  in  evolution  must  occur 
in  some  other  way. 

It  may  be  that  progressive  evolution  occurs  more  frequently  as  the  result 
of  changes  in  the  chromosomes  than  of  other  organelles.  But  the  present  hy- 
pothesis does  not  exclude  the  possibility  that  advances  in  evolution  can 
occur  by  ''progressive"  changes  in  the  composition  of  any  one  of  the  eternal 
organelles  such  as  the  nuclear  membrane  or  the  centrosome.  The  condition 
for  the  perpetuation  of  any  change  would  be  that  the  mutated  organelle 
could  be  provided  with  the  materials  necessary  for  its  continuance  by  the 
cell  as  a  whole  in  its  surrounding  environment  at  the  time  of  its  occurrence. 
On  this  hypothesis,  progressive  changes  in  evolution  are  not  confined  to  any 
single  cellular  component,  but  constitute  a  potential  of  every  component  of 
the  cell.  Although  progressive  changes  of  the  different  substances  compris- 
ing the  chromosome  may  not  occur  significantly  more  frequently  than 
changes  in  the  substances  making  up  the  other  organelles,  more  changes  may 
occur  in  the  chromosomes  in  toto  because  a  change  in  each  individual  com- 
ponent of  the  extraordinarily  heterogeneous  chromosome  registers  as  a  sepa- 
rate change. 


266  CARL  C.  LINDEGREN 

In  many  types  of  organisms  the  chromosomes  are  always  separated  by  the 
nuclear  membrane  from  the  cytoplasm.  The  mitochondria  (like  the  chromo- 
somes) are  relatively  non-homogeneous,  but  apparently  the  balance  of  their 
activities  is  not  so  critical  since  no  specific  devices  appear  to  be  required  to 
limit  their  reproduction  or  activity.  The  cytoplasm  is  probably  heterogene- 
ous also,  with  every  separate  eternal  component  having  the  same  continuity 
in  time  as  the  chromosomes.  However,  it  comprises  substances  transmitted  to 
the  daughter  cells  in  a  manner  which  is  apparently  subject  to  control  by  the 
environment,  and  this  may  constitute  the  basis  for  differentiation.  In  the 
germ  line,  the  entire  cytoplasmic  potential  must  be  maintained.  In  fact,  the 
main  function  of  the  germ  line  under  this  hypothesis  would  be  to  maintain 
an  intact  cytoplasm.  The  integrity  of  the  chromosomes  is  usually  provided 
for  in  either  the  somatic  or  the  germinal  tract.  Defects  in  the  extra  chromo- 
somal apparati  are  reconstituted  in  an  outcross,  thus  differentiating  so-called 
cytoplasmic  from  genie  inheritance. 


H.  H.  PLOUGH 

Amhersf  College 


Chapter  17 

Genetic  Implications  of 
Mutations  in  S.  Typhimurium* 


The  contribution  that  an  account  of  studies  in  bacterial  genetics  can  make 
to  the  problem  of  heterosis  must  be  indirect,  since  actual  sexual  or  other 
fusion  in  bacteria  has  not  been  observed  and  the  weight  of  evidence  is  against 
the  view.  Even  the  very  interesting  genetic  evidence  of  recombination  dis- 
covered by  Tatum  and  Lederberg  (1947)  in  the  K12  strain  of  the  colon 
bacillus,  and  now  being  developed  by  the  capable  studies  of  Lederberg  (1947, 
1949)  and  others,  is  still  susceptible  of  other  interpretations.  Diploid  strains, 
if  they  occur  at  all,  are  certainly  so  rare  as  to  be  unimportant  in  the  produc- 
tion of  hybrid  vigor  in  bacterial  populations. 

The  apphcations  of  bacterial  genetics  to  the  problem  of  heterosis  must  be 
rather  in  the  information  they  make  available  concerning  the  kinds  and  fre- 
quencies of  gene  mutations,  and  the  ways  in  which  they  interact  with  each 
other  within  populations.  It  has  been  generally  recognized  by  geneticists 
only  recently  that  the  bacteria  are  excellent  material  for  studies  of  these 
problems,  though  bacterial  mutation  was  first  mentioned  by  Massini  in  1907, 
and  distinctive  and  precise  food  requirements  for  bacterial  strains  have  been 
known  since  1913  (Hinselwood,  1946).  Studies  in  the  genetics  of  bacteria 
have,  of  course,  been  greatly  stimulated  by  the  pioneer  work  on  mutations 
in  fungi  by  Thom  and  Steinberg  (1939),  and  particularly  on  Neurospora  by 
Dodge,  by  Lindegren,  and  by  Beadle  (1949)  and  his  associates,  as  well  as  by 
the  important  work  on  yeast  as  presented  in  Dr.  Lindegren's  chapter. 
Long  before  the  currently  enlarging  wave  of  interest  in  bacteria  as  objects 
of  genetic  study,  Gowen  had  shown  that  mutations  of  the  same  order  of 
frequency  as  in  higher  plants  or  animals  were  induced  by  radiation  in  Phy  to- 

*  This  research  was  supported  bva  grant  from  the  Atomic  Energy  Commission,  Division 
of  Biology  and  Medicine  #AT (30-1) -930. 

267 


268  H.  H.  PLOUGH 

monas  (1945).  He  and  Zelle  had  indicated  the  genetic  basis  of  virulence  in 
Salmonella  (Zelle,  1942). 

ADVANTAGES  OF  SALMONELLA  FOR  GENETIC  STUDIES 

I  became  acquainted  with  the  Enterobacteriaceae  and  particularly  with 
the  pathogenic  forms  in  Zinsser's  laboratory  at  the  Columbia  Medical  School. 
My  own  realization  that  Salmonella  offered  excellent  material  for  studies  in 
microbial  genetics  was  heightened  when,  as  an  Army  bacteriologist  in  the 
Philippines,  I  had  to  diagnose  enteric  infections.  I  found  most  of  the  Salmo- 
nellas  which  Flexner  first  described  from  Manila  still  present  in  the  islands. 
More  than  140  strains  or  species  of  Salmonella  are  recognized  which  are  dis- 
tinguishable by  a  common  pattern  of  fermentation  reactions  (dextrose  and 
maltose- AG,  lactose  and  sucrose-negative,  citrate  and  H2S  positive).  Each 
one  has  been  shown  by  the  serological  studies  of  White  (1929),  Kaufmann 
(1944),  or  Edwards  and  Bruner  (1942)  to  have  a  very  precise  and  readily 
separable  antigenic  constitution. 

The  antigens  are  determined  by  agglutination  studies  using  serums  from 
different  rabbits  immunized  to  one  or  another  of  the  major  strains.  They 
fall  into  two  distinct  groups:  the  somatic  (0)  antigens  associated  with  the 
surface  protein  layers,  and  the  flagellar  (H)  antigens  determined  by  proteins 
of  the  flagella.  Each  of  these  groups  is  known  to  be  compound,  with 
some  twenty  separate  O  antigens — each  strain  may  carry  three  or  four 
(O)  antigens — and  eight  or  ten  different  specific  (H)  antigens  as  well  as  cer- 
tain alternative  and  non-specific  phases  of  the  latter.  Thus  each  strain  can 
be  shown  to  have  a  distinctive  and  readily  determinable  antigenic  constitu- 
tion (S.  typhimurium  is  I,  IV,  V,  XII — i,  1,  2,  3).  The  whole  group  naturally 
falls  into  a  tree-like  pattern  very  like  the  evolutionary  trees  made  for  fami- 
lies of  animals  or  plants  on  the  basis  of  structure. 

Tatum's  (1946)  discovery  that  mutagenic  agents  (including  radiation  and 
nitrogen  mustards)  could  induce  mutants  of  colon  bacteria  having  constant 
growth  factor  requirements  more  limited  than  the  parental  organism,  just 
as  with  Neurospora,  has  re-emphasized  the  one  gene-one  enzyme  hypothesis. 
It  has  strengthened  the  idea  of  bacterial  evolution  developed  by  Lwoff  (1943) 
that  the  parasitic  forms  have  been  derived  from  the  less  exacting  hetero- 
trophic organisms  by  successive  losses  of  synthetic  abilities.  Thus  it  gives 
added  meaning  to  the  tree-like  interrelationships  suggested  by  the  antigenic 
analyses. 

Soon  after  the  war  our  Amherst  group  entered  on  an  intensive  study  of 
induced  biochemical  and  antigenic  mutations  in  the  food  poisoning  organ- 
ism, Salmonella  typhimurium.  It  was  our  hope  that  this  organism  would 
prove  more  favorable  for  genetic  studies  than  E.  coli,  not  only  for  the  analysis 
of  the  mode  of  action  of  genes,  but  for  evidence  on  the  genetic  nature  of  type 
specificity,  virulence,  and  their  bearing  on  evolutionary  relationships. 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM  269 

METHODS  OF  INDUCING  AUXOTROPHIC  MUTATIONS 

The  strains  of  Salmonella  typhimuriimi  which  we  have  used  are  two:  519 
received  from  the  New  York  Salmonella  center  at  Beth  Israel  Hospital,  and 
533  (lie)  from  Gowen. 

Our  method  for  isolating  mutations  to  specific  food  or  growth  factor  re- 
quirements by  penicillin  screening  is  that  of  Lederberg  and  Zinder  (1948) 
and  of  Davis  (1949)  with  some  additions  of  our  own.  S.  typhimurium  is  a 
heterotrophic  organism  of  the  least  exacting  sort.  Cultures  will  grow  on  a 
basic  medium  containing  ammonium  sulphate,  sodium  chloride,  potassium 
phosphate  buffers,  with  traces  of  other  metallic  ions,  and  glucose  added  as 
an  energy  source.  Better  growth  is  obtained  with  a  supplementary  nitrogen 
source,  such  as  asparagin,  and  a  further  energy  source,  citrate,  but  these  are 
not  essential.  Thus  the  organism  synthesizes  all  its  own  food  comj)onents, 
coenzymes,  and  growth  factors,  as  well  as  the  enzymes  necessary  for  food 
and  energy  tranformations. 

Suspensions  are  subjected  to  radiation  by  X-rays  (up  to  100,000  roent- 
gens) or  ultraviolet  light  (up  to  3,600  ergs  per  mm.-),  and  are  then  transferred 
to  an  enriched  nutrient  broth  for  24  hours.  The  broth  stimulates  active 
division  of  all  organisms.  These  are  centrifuged  off,  washed,  and  reinoculated 
for  24  hours  into  the  basic  or  minimal  medium  containing  100  units  per  ml. 
of  penicillin.  This  stops  the  divisions,  and  progressively  kills  the  organisms 
which  divide  actively. 

These  organisms  which  penicillin  screens  out  are  called  prototrophic 
(Lederberg),  and  they  are,  of  course,  the  unchanged  originals.  Any  mutated 
organisms  w^hich  now  require  some  specific  nutrilite  will  not  divide  on  the 
basic  medium,  and  so  they  are  not  affected  by  penicillin.  These  are  now 
auxotrophic  organisms  (Davis),  and  they  are  isolated  by  plating  on  complete 
agar,  and  identified  by  paper  disc  inoculations  on  successive  plates  of  basic 
medium  with  single  nutrilites  added — amino  acids,  nucleic  acid  fractions,  or 
vitamins,  as  shown  in  Figures  17.1  and  17.2.  These  methods  are  described 
in  more  detail  by  Plough,  Young,  and  Grimm  (1950). 

AUXOTROPHIC  MUTATIONS  FROM  RADIATED  LINES 
I  shall  cite  only  one  set  of  isolations  from  such  a  radiation  experiment,  the 
data  for  which  are  given  in  Table  17.1.  Suspensions  from  an  unradiated  con- 
trol and  from  seven  successively  increased  X-radiation  dosages  were  run 
through  the  penicillin  screening,  and  500  auxotrophic  mutants  isolated.  Of 
these  a  total  of  459  were  recovered  and  their  specific  requirements  deter- 
mined. Although  the  control  had  been  derived  from  successive  single  colony 
isolations  within  3  days  of  the  tests,  still  5  per  cent  of  the  isolated  strains  were 
mutants — indicating  that  spontaneous  mutation  occurs  and  accumulates  in 
stock  strains. 

From  the  major  strain  used  (#533),  234  strains  out  of  the  459  isolated 


5   ML 


5    ML 


CENTRIFUGE 


24-HOUR  17  ML 

AGAR   SLANTS  NACL 


CELLS 
IN    NACL 


0  1     ML 


I     ML 


COMPLETE 
BROTH 

ISOLATIONS 


If 


3  ML 


10*  100  UNITS 

DILUTION      PENICILLIN 
IN    MIN.  MED. 


COMPLETE 
AGAR 


MINIMAL 
BROTH 


IN   3  ML 
OF   NACL 


COMPLETE 
BROTH 


Fig.  17.1 — -Diagram  showing  methods  for  the  production  of  radiation-induced  auxotrophic 
mutations  in  Salmonella  and  for  their  isolation  by  screening  through  minimal  medium 

containing  penicillin. 


PAPER  DISC  METHOD   FOR    TESTING   BIOCHEMICAL    MUTANTS 


TEST  STRAIN 

NEEDLE   TRANSFER 


24-HR  BROTH         DISTILLED    Hg  0 


PAPER  DISKS 


SALTS  +GLUCOSE  + 
AGAR  +  NUTRILITE 


Fig.  17.2 — Diagram  showing  method  for  determining  the  particular  nutrilite  required  by 
the  auxotrophs  isolated  as  in  Figure  17.1.  .\  series  of  Petri  plates  is  used,  each  containing 

a  different  test  substance. 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM 


271 


were  auxotrophic  mutants,  among  which  17  different  auxotroj^hic  mutants 
occur  according  to  tests  of  the  specific  nutrilite  required.  A  summary  of  these 
requirements  with  the  numbers  of  each  is  given  in  Table  17.2.  Tlae  most 
frequent  auxotroph  is  the  one  requiring  cysteine.  The  next  most  frequent  is 
the  histidine  auxotroph,  and  so  on  down  the  list  to  one  which  has  a  double 
requirement  of  both  valine  and  isoleucine  for  growth.  Only  two  auxotrophic 
mutants  require  substances  other  than  amino  acids.  One  must  be  supplied 

TABLE  17.1 

FREQUENCIES  OF  AUXOTROPHIC  MUTATIONS  IN  S.  TYPHIMURIUM 
AFTER  X-RADIATION  AND  PENICILLIN  SCREENING 


1 

X-Ray  Dosage 
and  Time 

2 

% 
Bacteria 
Surviving 

3 

Total 

No. 

Tests 

4 

Total 

No. 

Mutants 

5 

% 
Mutants 

6 

No. 
Different 
Mutants 

7 

% 
Different 

Mutants 

Strain  533 

I  Controls 

II  11,400  R4    min 

III  17,100  R6    min 

IV  22,800  R  8    min 

V  28,500  RIO  min 

VI  34,300  R  12  min 

VII  45,600  R  16  min 

VIII  57,000  R  20  min 

100 

40 

25 

14 
6 

1.5 
0.9 

135 
62 
86 
41 
25 
94 
99 
50 

7 

16 
18 
11 
19 
64 
72 
34 

5.1 
25.9 
20.9 
26.8 
76.0 
68.1 
72.6 
68.0 

3 
4 
4 
3 
6 
8 
17 
12 

2.2 

6.4 

4.7 

7.3 

24.0 

8.4 

17.2 

24.0 

Totals             

459 

234 

IX      II  +  III  +  IV 

189 
268 

45 
189 

23.8 
70.5 

11 
43 

5.8 

X       V+ VI 4-  VII 4- VIII. . . 

16.0 

Strain  519 
XI     45,600  R  16  min 

25 

100 

22 

22.0 

9 

9.0 

with  adenine,  and  others  (not  found  in  this  experiment)  must  have  either 
guanine  or  thiamin  in  the  medium. 

In  our  published  report  of  these  data  (Plough,  Young,  and  Grimm,  1950, 
Table  3)  we  listed  a  number  of  additional  strains  showing  alternative  re- 
quirements. Davis  (personal  communication)  retested  a  number  of  these 
and  found  them  to  be  mixtures  of  single  autotrophs.  We  have  just  completed 
an  extensive  recheck  of  all  strains  listed  originally  as  alternates,  and  now 
confirm  his  results  except  for  the  three  types  of  alternates  listed  in  Table 
17.2  (Plough,  Miller,  and  Berry,  1951). 

MUTATION  FREQUENCY  AND  X-RAY  DOSAGE 

One  of  the  most  interesting  results  of  this  experiment  is  the  clear  relation 
between  the  frequency  of  auxotrophic  mutants  and  the  X-ray  dosage.  This 
is  shown  in  Table  17.1,  column  5,  lines  II- VIII,  and  I  could  add  to  the  data 
from  other  experiments.  The  numbers  of  tests  vary  for  the  different  radiation 


272 


H.  H.  PLOUGH 


dosages,  and  some  of  the  values  are  less  significant  statistically,  but  the  per- 
centage of  mutants  is  significantly  higher  at  the  higher  dosages.  This  is  em- 
phasized by  lines  IX  and  X  where  the  sums  of  the  first  three  and  the  last 
four  values  are  compared.  The  same  conclusion  is  evident  from  inspection 
of  column  7  in  the  table,  where  the  numbers  of  different  mutants  at  the 
successive  dosages  are  shown.  Nearly  three  times  as  many  were  isolated  from 
the  upper  group  as  from  the  lower. 

TABLE  17.2 
KINDS  OF  AUXOTROPHIC  MUTATIONS  IN  S.  TYPHIMURIUM 


No. 

Strain  533 
Single  Amino  Acids 

No. 

Strain  519 
Single  Amino  Acids 

105 

55 

15 

14 

5 

5 

4 

2 

1 

Cysteine 

Histidine 

Leucine 

Proline 

Tyrosine 

Threonine 

Methionine 

Vahne 

Arginine 

8 

3 

3 

3 

1 

1 

1 

Histidine 

Cysteine 

Methionine 

Proline 

Leucine 

Tryptophane 

Phenylalanine 

5 

Nucleic  Acid  Fraction 
Adenine 

1 

4 

Multiple  Amino  Acids 
Valine  and  isoleucine 
Unanalyzed 

21 

1 

Alternative  A  niino  Acids 
Cysteine  or  methionine 
Tyrosine  or  tryptophane 

1 

Alternative  A  mino  Acids 
Cysteine  or  Methionine 

1 

Tyrosine  or  phenylalanine 

Line  XI  in  the  table  shows  the  result  of  one  radiated  series  made  on  a 
different  initial  strain,  #519.  Comparison  of  the  column  5  and  column  7 
totals  with  line  VII  above,  shows  that  this  strain  is  much  more  resistant  to 
radiation  than  is  strain  #533.  It  is  clear  that  comparisons  of  the  mutagenic 
effects  of  radiation  dosage  must  always  be  made  between  samples  from  the 
same  strain. 

The  data  in  Table  17.1,  column  5,  are  graphed  in  Figure  17.3.  Compari- 
son of  the  percentages  of  mutants  at  successive  dosages  shows  a  positive 
correlation,  though  rather  far  from  a  straight  line  curve.  As  the  penicillin 
screening  method  involves  a  24  hour  growth  in  complete  broth,  and  another 
24  hours  in  minimal  medium  with  penicillin,  it  might  be  expected  that  the 
final  percentage  of  mutant  strains  would  not  bear  the  direct  relation  to  dos- 
age shown  in  tests  of  mutations  produced  in  germ  cells  in  sexually  reproduc- 
ing organisms.  Indeed  Davis,  in  his  account  of  the  penicillin  screening  method 


GENETIC   IMPLICATIONS  OF  MUTATIONS  IN   S.  TYPHIMURIUM 


273 


as  used  in  E.  coli,  stated  "...  the  method  as  developed  so  far  does  not  appear 
to  yield  quantitative  survival  of  mutants."  Such  a  statement  assumes  that 
the  penicillin  screening  may  be  expected  to  be  complete,  which  in  fact  is  not 
true.  Rather  penicillin  acts,  as  do  all  antibiotics,  in  a  progressive  fashion  ac- 
cording to  a  typical  logarithmic  killing  curve.  If  two  or  more  mutant  cells 

PERCENT 
MUTATIONS 
100 


80 


60 


40 


20 


DATA    OF   TABLE    1 
A  =     I  -  VllI 
O  -     IX-  X 


o 


4        6        8       10       12       14       16 
MINUTES     OF     RADIATION 


20 


Fig.   17.3 — Graph  showing  the  relation  between  percentage  of  mutations  isolated  and 
X-ray  dosage  in  minutes  (2850  R  per  minute). 


appear  in  a  growing  wild  type  population,  they  will  increase  logarithmically 
and  form  smaller  less  numerous  clones.  As  the  penicillin  acts,  the  far  more 
numerous  parent  clones  will  be  logarithmically  reduced  in  numbers,  while 
the  mutant  clones  exposed  will  have  reached  a  level  which  may  be  main- 
tained during  the  24  hour  period  of  penicillin  action.  It  is  clear  that  if  a 
sample  is  taken,  and  plated  at  any  point  short  of  the  complete  killing  oflf  of 
the  wild  type,  we  may  expect  frequencies  showing  the  same  order  as  in  the 
original  population,  although  the  mutant  percentages  are  greatly  magnified. 


274 


H.  H.  PLOUGH 


An  actual  test  of  artificially  made  mixtures  of  the  parent  strain  and  one 
cysteine  requiring  mutant  as  screened  by  the  media  is  shown  in  Table  17.3. 
The  data  show  that  a  mixture  of  90  per  cent  wild  and  10  per  cent  mutant 
still  gives  a  greater  number  of  wild  survivors  after  penicillin  screening  than 
does  a  mixture  having  10  per  cent  wild  and  90  per  cent  mutant.  For  the 
actual  experiments  reported  in  Table  17.1  the  proportion  of  mutants  to  un- 
mutated  wild  type  even  after  24  hours  of  growth  in  complete  broth  is  one 
in  many  thousands,  rather  than  10  per  cent  to  90  per  cent.  So  it  seems  justi- 
fied to  consider  the  percentage  of  mutants  and  wild  type  as  an  index  of  muta- 

TABLE  17.3 

EFFECT  OF  GROWTH  IN  COMPLETE  MEDIUM  FOLLOWED 
BY  PENICILLIN  SCREENING  ON  ARTIFICIAL  MIX- 
TURES OF  CONTROL  (533)  AND  A  CYSTEINE  AUXO- 
TROPH  (533-169) 


Mixture 

Percentages 
Original  Mixture 

Percentages 

After  24  Hrs. 

IN  Broth 

Percentages 

After  Subsequent 

Screening 

533 

533-169 

533 

533-169 

533 

533-169 

A 

90 

10 

70 

30 

8 

92 

B 

50 

50 

33 

67 

2 

98 

C 

10 

90 

5 

95 

2 

98 

tion  frequency  in  comparing  X-ray  dosages.  The  trend  in  Figure  17.3  sug- 
gests a  sigmoid  curve  rather  than  a  straight  line  as  Hollaender  (1948)  has 
shown  for  ultraviolet  induced  visible  mutations  in  fungi.  Essentially  the 
same  interpretation  can  be  drawn  from  a  comparison  of  the  number  of  differ- 
ent mutations  found  at  the  successive  X-ray  dosages.  Much  more  extensive 
data  are  now  available  showing  the  relation  between  mutation  frequency 
and  both  X-radiation  and  ultraviolet  dosages  and  they  will  appear  in  another 
publication.  In  general  they  all  bear  out  the  conclusion  that  the  frequency 
of  auxotrophic  mutations  is  directly  correlated  with  radiation  dosage  as  is 
true  for  gene  mutation  in  other  organisms. 

A  rather  interesting  result  of  comparison  of  these  percentages  of  mutants 
present  after  penicillin  screening  is  that  the  most  frequent  class  changes  from 
the  lower  to  the  higher  dosages.  Thus  after  11,000  roentgens,  a  cysteine  auxo- 
troph  is  the  most  frequent,  while  after  57,000  r  it  is  a  histidine  requirer. 
Perhaps  we  are  dealing  with  a  specific  effect  of  dosage  or  conceivably  with 
a  differential  effect  of  wave  length,  but  until  the  complex  nature  of  the  cys- 
teine mutants  are  more  fully  understood  it  is  unwise  to  attempt  too  definite 
an  interpretation. 

RECOMBINATION  TESTS  IN  SALMONELLA 

Much  interest  has  been  excited  among  geneticists  as  well  as  bacteriologists 
by  Lederberg's  proof  that  mixtures  of  multiple  mutant  stocks  of  the  K12 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM  275 

strain  of  E.  coli  give  rise  to  new  strains  having  the  auxotrophic  mutants  in 
new  combinations.  These  initial  observations  have  been  repeated  in  differ- 
ent combinations  and  amply  confirmed  by  the  observations  of  many  other 
investigators.  As  Lederberg  has  suggested,  these  results  are  most  reasonably 
interpreted  as  due  to  bacterial  union  like  a  sexual  fusion  of  gametes,  fol- 
lowed by  an  immediate  reduction  process  involving  segregation  and  genetic 
recombination,  suggesting  linkage  in  a  single  chromosome  system.  More 
recently  Lederberg  (1949)  has  found  evidence  of  what  appears  to  be  a  diploid 
strain  which  gives  highly  aberrant  segregation  ratios.  These  require  assump- 
tions of  such  an  extremely  complex  and  involved  type  of  chromosome 
interchange  that  it  becomes  questionable  whether  some  other  explanation  is 
not  after  all  more  probable. 

In  S.  typ/iinmrium  we  now  have  more  mutant  strains  carrying  single 
auxotrophic  genes  or  multiple  combinations  of  these  than  in  any  other 
bacterial  species  except  E.  coli.  This  makes  it  especially  important  to  test  the 
theory  with  our  strains.  Accordingly  Miss  Marie  McCarthy  has  been  mixing 
these  in  varying  combinations,  and  then  plating  out  in  heavy  suspensions  on 
base  medium  supplemented  so  as  to  show  up  the  transfer  of  one  or  more  re- 
quirements from  one  to  the  other  original  combination. 

Although  more  than  a  hundred  such  tests  have  been  made  and  carefully 
checked,  the  results  have  been  unequivocally  negative  until  very  recently. 
This  work  will  be  reported  in  detail  in  a  later  publication,  but  I  will  describe 
it  briefly  here.  Multiple  strain  #519-38-94-41  requiring  tryptophane,  me- 
thionine, and  histidine  was  mixed  with  #533-486-96-85  requiring  leucine, 
threonine,  and  arginine.  On  plating  in  appropriate  media  it  was  found  that 
in  addition  to  the  original  parental  combinations  several  colonies  each  gave 
strains  requiring  two  new  sets  of  requirements.  Recombination  No.  1  re- 
quired tryptophane,  leucine,  and  threonine.  Recombination  No.  2  needed  all 
six  amino  acids:  tryptophane,  methionine,  histidine,  leucine,  threonine,  and 
arginine.  These  new  stocks  have  been  retested,  and  there  can  be  no  question 
of  the  fact  that  we  have  here  two  recombinations  of  the  original  stocks  used. 
Other  recombinations  have  now  appeared  but  reciprocal  classes  are  never 
found.  Thus  we  have  in  Salmonella  confirmation  of  the  recombination  results 
found  by  Lederberg  in  the  K12  strain  of  E.  coli.  In  view  of  the  irregularity  of 
such  results  both  in  E.  coli  and  in  Salmonella,  it  would  seem  wise  to  suggest 
that  some  alternative  explanation  may  yet  prove  to  be  more  satisfactory  than 
recombination  or  chromosomal  crossing-over. 

BIOCHEMICAL  STUDIES  OF  AUXOTROPHIC  MUTANTS 

The  Neurospora  studies  of  Eeadle  and  his  associates  as  well  as  those  of 
Lindegren  (1949)  on  yeast  have  made  it  evident  that  in  studying  the  action 
of  auxotrophic  mutants  we  are  many  steps  closer  to  the  initial  determinative 
activities  of  the  genes  themselves  than  is  ordinarily  true  for  characters  in  the 
higher  plants  and  animals.  When  a  series  of  auxotrophic  genes  can  be  shown 


276 


H.  H.  PLOUGH 


to  block  successive  steps  in  the  syntheses  of  particular  amino  acids  or  vita- 
mins or  more  complex  products,  the  one  gene-one  enzyme  hypothesis  offers 
the  most  satisfactory  preliminary  explanation,  even  though  the  presence  of 
the  particular  enzyme  as  a  gene  product  has  not  been  demonstrated.  Each 
set  of  auxotrophic  mutants  offers  data  on  the  chain  of  synthetic  processes  to 
some  essential  substance,  and  thus  becomes  a  challenging  biochemical  prob- 
lem. It  is  significant  that  many  of  those  already  studied  in  the  fungi  have 
also  been  uncovered  in  E.  coli,  but  every  organism  shows  individual  differ- 
ences. So  far  in  Salmonella  we  have  investigated  the  biochemical  steps  in 
only  two  such  series  of  auxotrophs,  but  many  others  await  study  especially 
as  new  mutants  are  added. 

TABLE  17.4 

UTILIZATION  OF  SULPHUR  COMPOUNDS  BY  VARIOUS 
AUXOTROPHS  OF  5.  TYPHIMURIUM 


Strain 

NaiSOi 

NajSiOs 

NajS 

Cysteine 

Cysta- 
thionine 

Methio- 
nine 

Block 
in  Fig.  17.4 

1.  Original  533 

2.  533-575 

+ 



+ 
+ 

+ 
+ 
+ 

I     1   -f  +  +  + 

+ 
+ 
+ 
+ 
+ 

+ 

+ 

+ 
+ 

None 
7-1-2 

3.  533-526 

7 

4.  533-452 

5-h2 

5.  533-P249 

6.  533-535 

4 
2 

The  first  of  these  sets  of  interacting  synthetic  steps  which  we  have  studied 
is  the  cysteine-methionine  auxotroph  series.  These  mutants  fall  into  many 
of  the  same  gradations  described  by  Lampen,  Roepke,  and  Jones  (1947)  for 
E.  coli,  by  Emerson  (1950)  for  Neurospora,  and  by  Teas  (1950)  for  B. 
subtilis.  We  have  tested  all  of  the  apparent  cysteine  or  methionine  requirers 
for  their  ability  to  reduce  inorganic  sulfur  compounds  as  well  as  to  utilize 
organic  precursors  of  methionine.  The  wild  type  strains  can  reduce  sulphate, 
sulphite,  or  sulfide,  and  can  grow  with  no  other  source  of  S.  It  has  been 
shown,  however,  that  none  of  the  apparent  cysteine  requirers  can  reduce 
sulphate,  but  some  can  reduce  sulphite  and  some  sulfide.  Many,  however, 
must  have  cysteine  or  cystathionine  (kindly  supplied  by  Dr.  Cowie)  and 
others  require  methionine  as  such. 

A  summary  of  representative  mutants  isolated  as  cysteine  or  methionine 
requirers  and  their  abilities  to  grow  on  various  compounds  as  the  sole  source 
of  S  is  given  in  Table  17.4.  This  can  be  visualized  as  in  Figure  17.4  in  terms 
of  a  succession  of  steps,  each  catalyzed  by  an  enzyme  controlled  by  a  gene 
which  is  inactivated  by  the  mutation  numbered  in  parentheses.  Such  a 
straight  line  series  appears  to  run  in  the  direction  of  the  arrows  from  sulphate 
to  protein.  When  a  mutation  occurs,  as  at  (5),  it  must  be  assumed  that  growth 
requirements  will  be  satisfied  by  any  compound  succeeding  the  break  in  the 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM  277 

synthetic  chain,  unless  a  second  mutation  has  occurred.  This  does  not  hold 
for  methionine  which  cannot  be  utilized  in  mutants  #2  and  4  (Table  17.4). 
Such  a  result  suggests  that  cysteine  is  enzyme  controlled  through  a  gene 
which  is  inactivated  by  the  mutation  numbered  in  parentheses.  Cysteine  is 
ordinarily  made  from  methionine  (as  has  been  shown  for  the  mammal)  and  so 
the  reverse  dotted  arrow  marked  (1)  is  shown  in  the  figure.  It  is  hardly  likely 
that  a  second  mutation  is  indicated  for  the  mutants  cited  as  showing  two 
blocks,  but  rather  that  certain  mutations  cause  inhibition  of  more  than  one 
enzyme  system.  A  more  comprehensive  scheme  for  the  cysteine-methionine 
synthesis  based  on  the  Neurospora  work  has  been  given  by  Emerson  (1950). 
It  is  certain  that  more  is  involved  in  the  series  of  reactions  shown  in  Figure 
17.4  than  the  furnishing  of  essential  sulphur  for  cysteine  and  methionine. 

^  Protein 

(7)                         (6)                    (5) 
SO4 »>  SO3  *►  S  Cysteine 

HOMOSERINE 
(1)/  \    (4) 

y  Cystathionine 

?  ^  d)  \ 

Protein  -< Methionine  < Homocysteine 


Serine 

Fig.  17.4 — Possible  chain  of  reactions  involving  sulphur-containing  compounds.  (Mutant 
blocks  indicated  by  numbers  in  parenthesis.) 

Sulphate,  sulphite,  and  suliide,  as  well  as  cysteine  itself,  may  act  as  H 
acceptors,  cooperating  with  dehydrogenases  involved  in  the  respiratory  or 
energy  producing  activities  of  the  organism.  That  the  organism  reduces  more 
sulphate  than  is  necessary  for  the  S  required  in  the  amino  acids  is  indicated 
by  the  fact  that  Salmonella  forms  a  readily  testable  excess  of  H2S.  We  are 
attempting  to  trace  the  course  of  the  sulphur  by  the  use  of  the  radioactive 
isotope  S^''.  Last  summer  Dr.  T.  P.  Ting  and  the  writer  were  able  to  show  that 
(NH4)2S*04  is  taken  up  by  the  wild  type  533  organisms  and  not  at  all  by  a 
cysteine  requiring  mutant,  thus  confirming  our  growth  tests  (see  also  Cowie, 
Bolton,  and  Sands,  1950). 

We  hope  to  continue  this  work  using  labeled  sulphur  in  sodium  sulfide  or 
barium  sulfide,  which  should  be  utilized  by  wild  and  mutants  number  (7), 
(6),  (5)  (Fig.  17.4).  Finally  it  should  be  possible  to  determine  by  quanti- 
tative tests  how  much  S^*  is  combined  into  bacterial  protein  and  how  much 
passed  out  in  HoS.  Comparisons  between  different  strains  in  oxygen  utiliza- 
tion are  being  made  with  the  W'arburg  respirometer.  As  already  shown  in 


278 


H.  H.  PLOUGH 


Table  17.3,  some  cysteine  requiring  strains  will  overgrow  the  parent  and  this 
may  be  due  to  differences  in  energy  requirements. 

The  second  set  of  steps  in  synthesis  being  studied  concerns  the  adenine 
requirer.  Here  we  appear  to  have  rather  more  definite  information  than  was 
described  by  Guthrie  (1949)  for  the  purine  auxotrophs  of  E.  coli.  It  has  been 
shown  that  the  Salmonella  auxotroph  utilizes  adenine  and  hypoxanthine,  but 
not  guanine  and  xanthine.  Of  the  nucleosides  and  nucleotides  only  adenosine 
and  adenylic  acid  are  used,  and  much  more  of  the  latter  is  required  for  com- 
parable growth  than  of  adenine.  Thus  it  appears  that  in  purine  metabolism, 
Salmonella  and  an  animal  like  Tetrahymena  (Kidder  and  Dewey,  1948) 
show  almost  opposite  requirements,  for  the  bacteria  do  not  convert  adenine 
to  guanine.  Preliminary  studies  by  Mrs.  Helen  Y.  Miller  demonstrate  a  spar- 
ing action  for  adenine  utilization  by  the  amino  acid  histidine.  This  suggests 


N==C-NH2 


HC— NH 


HC— NH 


H-C 


C-NH 


N 


— C-N 

Adenine 


^ 


CH 


C— N 
CH2 

o=c 


^ 


CH 


-^- 


C — N 
CH2 
H2N-CH 


/- 


:cH 


COOH 

Imidazole  Pyruvic  Acid 

+ 
Pyridoxamine 


COOH 
Histidine 


Fig.  17.5 — A  Possible  relation  of  Adenine  to  Histidine  synthesis  (after  Broquist  and  Snell). 


that  for  this  organism  as  with  Lactobacillus  (Broquist  and  Snell,  1949)  the 
purine  is  a  precursor  of  histidine,  probably  by  the  utilization  of  the  imid- 
azole ring  through  pyruvic  acid,  and  the  transaminating  action  of  pyridoxa- 
mine (Figure  17.5).  While  these  facts  have  been  revealed  by  a  study  of  the 
adenine  mutant  alone,  further  gene  changes  and  their  reactions  with  the 
histidine  auxotrophs  already  available  should  help  clarify  some  of  the  inter- 
actions of  purines  and  amino  acids  in  the  bacterial  cell. 

ALTERATION  OF  ANTIGENIC  SPECIFICITY 

The  auxotrophic  mutations  reveal  a  series  of  biochemical  steps  or  trans- 
formations common  to  whole  groups  of  organisms.  Antigenic  analysis,  on  the 
other  hand,  has  revealed  precise  specific  or  strain  differences  which  are  as 
distinctive  as  the  form  or  structural  differences  of  complex  animals  and 
plants.  This  has  been  clearly  demonstrated  by  the  blood  group  analysis 
presented  in  the  studies  of  Irwin  and  his  colleagues.  The  specificity  is  no  less 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM  279 

sharp  in  the  antigenic  analyses  of  Salmonella.  The  tree-like  relationship 
which  they  suggest  was  our  chief  stimulus  to  a  study  of  bacterial  genetics  in 
this  organism. 

Preliminary  tests  of  all  of  the  auxotrophic  mutants  made  by  Miss  Dorothy 
Farley  show  that  they  are  unchanged  antigenically.  Not  only  the  specific 
antigens,  but  the  agglutination  titers  are  the  same  as  the  original  strains. 
This  has  been  confirmed  by  reciprocal  absorption  tests,  as  well  as  by  precipi- 
tation, and  inhibition  of  agglutination  using  supernatants  from  boiled  cul- 
tures. Thus  it  appears  that  the  loss  of  ability  to  synthesize  a  particular  amino 
acid  in  no  way  alters  the  antigenic  configuration.  Apparently  if  proteins  are 
formed  at  all  they  take  on  the  antigenic  configuration  of  the  cytoplasm  al- 
ready there.  The  auxotrophic  mutants  and  the  antigenic  patterns  fall  into 
two  quite  independent  systems  so  far  as  present  evidence  goes.  This  seems 
to  be  true  also  for  variations  in  or  loss  of  virulence.  The  relation  of  the  auxo- 
trophic mutants  to  virulence  for  mice  is  being  studied  in  detail  by  Gowen 
and  his  associates  and  will  be  reported  separately,  but  so  far  at  least  it  ap- 
pears that  there  is  no  relation  between  virulence  and  the  biochemical  re- 
quirements of  the  strain. 

It  was  originally  and  is  still  our  hope  to  be  able  to  induce  antigenic 
variants  by  radiation,  but  so  far  such  attempts  have  given  negative  results. 
We  have  inoculated  radiated  suspensions  into  one  end  of  U  tubes  of  semi- 
solid agar  containing  low  concentrations  of  O  serum  from  a  rabbit  im- 
munized against  the  specific  strain,  and  the  organisms  grow  through  the 
medium.  When  agar  containing  specific  H  serum  is  used,  however,  the 
organisms  grow  only  at  the  site  of  inoculation.  If  antigenic  mutants  had 
occurred  we  would  expect  that  the  homologous  serum  would  act  as  a  screen 
to  block  off  the  original  and  let  the  mutants  through,  just  as  the  penicillin 
does  for  the  auxotrophs.  The  result  simply  means  that  we  have  not  found  any 
antigenic  mutants  following  radiation.  Perhaps  we  should  not  expect  any. 
Antigenic  mutants  have  been  induced  in  several  bacteria  by  other  meth- 
ods, especially  by  McCarty  (1946)  in  the  pneumococcus,  by  Bruner  and 
Edwards  (1947)  in  Salmonella,  and  by  Boivin  (1947)  in  E.  coU.  The  pneu- 
mococcus method  is  not  applicable  to  Salmonella,  and  the  Boivin  method  in- 
volving exposure  of  the  organism  to  autolysates  of  rough  variants  of  other 
strains  gives  negative  results.  Tests  using  similar  culture  filtrates  have 
been  unsuccessful  in  altering  the  antigenic  constitution  of  our  organism.  On 
the  other  hand,  Miss  Farley  has  made  use  of  the  Edwards  technique  of  grow- 
ing an  auxotrophic  mutant  in  a  semi-solid  medium  containing  homologous  O 
serum  previously  absorbed  with  a  related  organism  which  lacked  one  of  the 
major  antigens,  XII  (and  in  another  case  lacked  V  but  carried  an  additional 
antigen  XXVII).  By  this  method  two  successful  transformations  of  type 
have  been  secured  out  of  several  tried.  Both  of  these  transformations  were 
performed  on  an  auxotrophic  mutant  (519-PlO)  requiring  histidine. 


280  H.  H.  PLOUGH 

Preliminary  tests  showed  that  these  strains  were  antigenically  similar  and 
gave  the  same  agglutination  titer  with  homologous  serum  as  the  parental 
wild  types — (I)  IV,  V,  XII,  for  the  O  antigens.  The  parentheses  indicate  that 
(I)  is  very  weak  or  absent.  The  first  case  is  typical.  Specific  serum  from  ani- 
mals immunized  by  #519  was  absorbed  with  a  suspension  of  organisms  of 
#527,  an  unnamed  strain  known  to  have  0  antigens  IV,  V  only.  After  it  was 
passed  through  semisolid  agar  containing  the  absorbed  serum  now  carrying 
XII  antibodies  only,  519-PlO  was  retested  and  shown  now  to  give  agglutina- 
tion at  a  very  low  titer  (1/320  instead  of  1/10,000)  compared  with  the  origi- 
nal. Further  testing  has  demonstrated  that  this  strain  retains  the  two  major 
0  antigens  (IV  and  V),  but  has  lost  XII.  Thus  it  has  been  transformed  to  IV, 
V  like  strain  #527.  Further  tests  on  differential  media  prove  that  the  strain 
is  unchanged  as  an  auxotrophic  mutant,  and  still  cannot  grow  unless  the 
medium  contains  histidine  (519-10). 

In  the  other  case  519  0  serum  was  absorbed  by  S.  schleissheim  (V,  XII, 
XXVII).  The  mutant  after  growing  through  the  absorbed  serum  failed  to 
agglutinate  in  XII  serum,  and  had  a  higher  titer  in  XXVII  than  S.  schleiss- 
heim. Thus  the  changed  mutant  has  lost  XII  and  taken  on  antigen  XXVII. 
It  still  retains  its  histidine  requirement. 

Thus  we  have  two  independent  cases  of  the  alteration  of  antigenic  speci- 
ficity by  the  Edwards  method  of  passage  through  specific  serum.  Here  again 
the  evidence  indicated  no  relation  between  antigenic  configuration  and  the 
biochemical  requirements.  We  are  now  exposing  these  antigenically  altered 
strains  to  further  radiation  with  the  idea  of  building  up  multiple  auxotrophic 
stocks  combining  the  two  major  systems  of  mutations.  These  can  then  be 
used  for  more  conclusive  tests  of  possible  fusion  and  recombination.  How- 
ever, this  demonstration  that  antigenic  mutants  can  be  induced  by  specific 
serum  adds  to  the  possibility  that  mutual  interaction  of  genes  or  gene 
products  between  organisms  in  mixtures  may  give  a  more  acceptable 
explanation  of  the  recorded  cases  of  recombination  in  bacteria,  than  does  one 
based  on  genetic  analogies  with  higher  forms. 

SUMMARY 

An  account  has  been  given  of  the  results  of  X-radiation  of  suspensions  of 
the  two  strains  of  Salmonella  lyphimurium,  and  the  isolation  of  strains  with 
specific  nutrilite  requirements  (auxotrophic  mutants).  These  strains  are  iso- 
lated by  the  Davis-Lederberg  method  of  growth  for  twenty-four  hours  in 
enriched  broth,  followed  by  twenty-four  hours  in  minimal  broth  containing 
100  units  per  ml.  of  penicillin.  The  method  screens  out  the  unmutated  organ- 
isms according  to  a  logarithmic  survival  curve,  and  preserves  the  mutant 
bacteria. 

Successive  tests  show  a  relation  between  X-ray  dosage  and  the  percentage 
of  recovered  auxotrophic  mutants,  and  also  between  dosage  and  the  number 
of  different  mutants. 


GENETIC  IMPLICATIONS  OF  MUTATIONS  IN  S.  TYPHIMURIUM  281 

In  all,  249  separate  auxotrophic  mutants,  of  which  20  are  different,  were 
isolated  out  of  459  tests.  Most  of  these  showed  requirements  for  single  amino 
acids,  but  a  few  required  the  purine  base  adenine,  and  others  showed  alterna- 
tive, and  a  small  number,  multiple  requirements. 

A  large  number  of  tests  involving  growth  of  multiple  mutant  stocks  in 
mixtures  followed  by  re-isolations  have  been  made  to  test  for  possible  fusion 
and  recombination  as  reported  by  Lederberg  and  others  in  the  K  12  strain 
of  E.  coli.  Recombination  has  been  found  but  it  is  unlike  that  in  sexually 
reproducing  organisms. 

Detailed  studies  of  the  different  auxotrophs  requiring  cysteine  or  me- 
thionine show  a  step-like  series  beginning  with  loss  of  ability  to  reduce  in- 
organic sulphate,  and  continuing  to  the  loss  of  ability  to  form  methionine. 
Many  of  these  mutational  steps  are  explainable  as  due  to  the  inactivation  of  a 
specific  enzyme,  but  several  require  a  complex  pattern  of  chemical  interac- 
tions. 

Similar  studies  of  the  adenine  auxotroph  suggest  that  adenine  may  be  a 
source  of  histidine. 

Tests  have  been  made  to  determine  if  antigenic  specificity  can  be  altered 
by  radiation,  with  negative  results.  However,  an  auxotrophic  mutant  has 
been  antigenically  altered  in  two  difTerent  cases  by  the  Edwards  technique  of 
passing  through  absorbed  immune  serum.  In  each  case,  one  of  the  0  antigens 
was  removed,  and  in  one  case  another  0  antigen  was  added.  In  both  cases  the 
biochemical  requirement  of  histidine  was  retained. 

It  appears  that  the  auxotrophic  and  antigenic  series  represent  two  quite 
different  and  unrelated  sets  of  mutations. 


JAMES  F.  CROW 

Universify  of  Wisconsin 


Chapter  1 8 

Dominance 
and  Overdominance 


Since  the  first  attempts  to  explain  hybrid  vigor  and  the  deleterious  effects 
of  inbreeding  in  Mendelian  terms,  there  have  been  two  principal  hypotheses. 
Both  were  advanced  early,  and  though  each  has  had  its  ups  and  downs  in 
popularity,  both  have  persisted  to  the  present  time.  The  first  hypothesis  is 
based  on  the  observed  correlation  between  dominance  and  beneficial  effect 
(or  recessiveness  and  detrimental  effect).  Inbreeding  uncovers  deleterious 
recessives,  and  typically  results  in  deterioration. 

With  hybridization,  some  of  the  detrimental  recessives  brought  into  the 
hybrid  zygote  by  one  parent  are  rendered  ineffective  by  their  dominant 
alleles  from  the  other,  and  an  increase  in  vigor  is  the  result.  If  the  number 
of  factors  is  large,  or  if  there  is  linkage,  the  probability  becomes  exceedingly 
small  of  a  single  inbred  line  becoming  homozygous  for  only  the  dominant 
beneficial  factors.  Consequently,  there  should  be  a  consistent  decrease  in 
vigor  with  inbreeding,  and  recovery  with  hybridization.  This  idea  has  been 
called  the  dominance  or  the  dominance  of  linked  genes  hypothesis. 

The  alternative  theory  assumes  that  there  is  something  about  hybridity 
per  se  that  contributes  to  vigor.  In  Mendelian  terms  this  means  that  there 
are  loci  at  which  the  heterozygote  is  superior  to  either  homozygote,  and  that 
there  is  increased  vigor  in  proportion  to  the  amount  of  heterozygosis.  This 
idea  has  been  called  stimulation  of  heterozygosis,  super-dominance,  over- 
dominance,  single  gene  heterosis,  cumulative  action  of  divergent  alleles,  and 
simply  heterosis. 

In  accordance  with  the  title  of  this  discussion  I  shall  use  the  words  domi- 
nance and  overdominance  for  the  two  hypotheses.  This  leaves  the  word 
heterosis  free  for  more  general  use  as  a  synonym  for  hybrid  vigor  (Shull,  1948). 

*  Paper  No.  434  from  the  Department  of  Genetics,  University  of  Wisconsin. 

282 


DOMINANCE  AND  OVERDOMINANCE  283 

In  most  situations,  the  hypotheses  of  dominance  and  overdominance  lead 
to  the  same  expectations.  In  either  case  there  is  a  decrease  of  vigor  on  in- 
breeding and  a  gain  on  outcrossing.  Wright  (1922c)  has  shown  that  with  the 
dominance  hypothesis  the  decline  in  vigor  is  proportional  to  the  decrease  in 
heterozygosis,  regardless  of  the  relative  number  of  dominant  and  recessive 
genes  and  of  the  degree  of  dominance.  The  same  decline  in  vigor  with  de- 
creasing heterozygosity  is  true  with  overdominance. 

It  is  usually  impossible  in  a  breeding  experiment  to  differentiate  between 
true  overdominance  in  a  pair  of  alleles,  and  pseudo-overdominance  due  to 
the  effects  of  two  pairs  of  alleles  closely  linked  in  the  repulsion  phase.  Only 
in  special  circumstances,  such  as  when  a  mutation  has  recently  occurred  in 
an  isogenic  stock,  can  the  experimenter  be  reasonably  certain  that  the  effect 
is  due  to  a  single  allelic  difference.  Furthermore,  there  is  the  possibility  of 
heterosis  due  to  borderline  situations,  such  as  might  arise  in  pseudoalleles 
with  a  position  effect,  which  could  not  even  theoretically  be  classified  as  due 
to  dominance  of  linked  genes  or  overdominance.  Finally,  it  should  be  noted 
that  the  various  hypotheses  may  not  be  equally  important  in  all  situations. 
For  example,  it  is  reasonable  to  expect  that  overdominance  would  be  more 
important  in  determining  differences  between  inbred  lines  of  corn  pre- 
viously selected  for  general  combining  ability  than  in  lines  not  so  selected. 

If  the  two  hypotheses  are  not  mutually  exclusive,  neither  are  they  col- 
lectively exhaustive.  There  is  no  reason  to  think  that  multiple  factors  are 
any  less  complex  in  their  interactions  than  factors  concerned  with  qualitative 
differences.  With  the  number  of  genes  involved  in  heterosis,  and  with  the 
complexity  of  interactions  known  to  exist  in  cases  where  individual  gene 
effects  have  been  isolated  and  studied,  there  must  surely  be  all  sorts  of  com- 
plex interactions  in  heterosis.  Therefore  no  single  theory  can  be  expected  to 
account  for  the  entire  effects  of  heterosis.  Although  it  is  difficult  to  separate 
by  statistical  methods  the  effects  of  dominance  and  epistasis,  it  may  be 
possible  to  construct  simple  models  which  are  of  some  utility. 

DOMINANCE 

Davenport  (1908)  was  the  first  to  point  out  the  now  well-recognized  fact 
that  in  most  cases  the  dominant  character  is  beneficial  to  the  organism  pos- 
sessing it,  while  the  recessive  has  a  weakening  effect.  He  noted  that  this  could 
help  explain  the  degeneration  that  usually  follows  inbreeding.  Davenport 
was  thinking  of  relatively  few  factors  with  individually  large  effects,  whereas 
at  present,  more  emphasis  is  given  to  multiple  factors.  But  he  was  close  to 
the  ideas  now  held. 

Keeble  and  Pellew  (1910)  found  that  hybrids  between  two  pure  varieties 
of  peas  were  taller  than  either  parent.  In  this  case,  two  different  dominant 
factors  were  involved — one  resulting  in  longer  internodes  and  the  other  in- 


284  JAMES  F.  CROW 

creasing  their  number.  Here  only  two  gene  pairs  were  involved,  but  it  was 
mentioned  that  similar  systems  might  hold  for  more  complex  cases. 

A  more  general  development  of  the  dominance  hypothesis  was  given  dur- 
ing the  same  year  by  Bruce  (1910).  He  designated  the  frequencies  of  domi- 
nant and  recessive  alleles  as  p  and  q  in  one  breed  and  P  and  Q  in  the  other. 
The  array  of  individuals  in  the  two  groups  will  then  be  {p'-DD  -\-  IpqDR  -\- 
q^RRY  and  {PWD  +  2PQDR  +  Q-RRY,  where  D  and  R  are  the  dominant 
and  recessive  alleles  and  n  is  the  number  of  factor  pairs  involved.'  If  these 
two  populations  are  crossed,  the  mean  number  of  homozygous  recessive  loci 
is  nqQ,  whereas  the  average  number  for  the  two  parent  populations  is 
n{q-  -f  Q~)/2.  The  former  is  the  geometric  mean  of  the  two  parental  recessive 
genotype  frequencies  while  the  latter  is  the  arithmetic  mean.  Since  the  geo- 
metric mean  is  always  less  than  the  arithmetic,  the  number  of  homozygous 
recessive  loci  will  always  be  less  in  the  hybrid  population  than  the  mean 
number  in  the  two  parent  populations.  If  either  or  both  the  parent  popula- 
tions are  inbred  the  decrease  will  be  greater. 

Bruce  then  said: 

If,  now,  it  be  assumed  that  dominance  is  positively  correlated  with  vigor,  we  have  the 
final  result  that  the  crossing  of  two  pure  breeds  produces  a  mean  vigor  greater  than  the  col- 
lective mean  vigor  of  the  parent  breeds.  ...  I  am  aware  that  there  is  no  experimental  evi- 
dence to  justify  the  assumption  that  dominance  is  correlated  with  a  "blending"  character 
like  vigor;  but  the  hypothesis  is  not  an  extravagant  one,  and  may  pass  until  a  better  takes 
the  field. 

The  average  proportion  of  recessive  homozygotes  in  the  parents,  which  is 
(9^  +  '2")/2,  may  be  rewritten  as  qQ -{-  {q  —  Q)~/2.  This  is  always  larger 
than  qQ,  the  proportion  in  the  hybrid,  unless  q  and  Q  are  equal.  Although 
Bruce  didn't  mention  this,  after  one  generation  of  random  mating  the  propor- 
tion of  recessives  in  the  hybrid  population  becomes  {q  +  QY/'^  =  (lQ'\~ 
(q  —  QY/4-,  which  shows  that  half  the  gain  in  vigor  is  lost  as  soon  as  ran- 
dom mating  begins. 

Bruce  concentrated  his  attention  on  the  decrease  of  homozygous  reces- 
sive loci  in  the  hybrid,  and  postulated  a  correlation  between  recessiveness 
and  deleterious  effect.  He  could  have  used  the  same  algebraic  procedures  to 
show  that  crossing  produces  an  increase  in  heterozygous  loci,  and  thus  based 
a  theory  of  hybrid  vigor  on  overdominance.  He  showed  remarkable  foresight 
in  choosing  the  former,  at  a  time  when  he  had  no  evidence  of  a  correlation 
between  dominance  and  beneficial  effect. 

1.  The  notation  used  by  Bruce  implies  equal  frequency  of  dominant  and  recessive 
alleles  at  all  loci.  This  assumption  is  not  at  all  necessary  for  the  argument,  and  I  think 
that  what  Bruce  really  meant  was 


Yl  {p]DD-^2p,q,DR-^qfRR). 


DOMINANCE  AND  OVERDOMINANCE  285 

Objections  to  the  dominance  hypothesis  were  made  largely  on  two 
grounds.  First,  if  vigor  is  not  a  product  of  heterozygosity  as  such,  it  should 
be  possible  by  selection  to  obtain  individuals  which  are  homozygous  for  all 
the  beneficial  dominant  factors,  and  hence  have  the  same  vigor  as  hybrids. 
Secondly,  in  the  Fo  of  a  cross  between  two  inbred  strains  there  should  be  a 
skew  distribution  of  the  trait  being  measured — since  the  dominant  and  re- 
cessive loci  would  be  distributed  according  to  the  expansion  of  (3/4  +  1/4)", 
where  n  is  the  number  of  factors. 

These  objections  were  largely  removed  when  Jones  (1917)  pointed  out 
that,  with  linkage,  the  consequences  of  the  dominance  hypothesis  were 
much  closer  to  those  postulating  superior  heterozygotes.  If  a  detrimental 
recessive  were  linked  with  a  favorable  dominant,  the  heterozygous  chromo- 
some would  be  superior  to  both  homozygotes,  and  the  linked  combination 
might  not  break  up  readily.  Later,  Collins  (1921)  showed  that  with  a  large 
number  of  factors,  regardless  of  linkage,  the  skew  distribution  disappears. 
The  probability  of  getting  all  the  beneficial  dominants  into  one  homozygous 
strain  becomes  vanishingly  small,  so  the  objections  hold  only  if  a  small  num- 
ber of  factors  is  assumed. 

Most  of  the  mutations  known  in  Drosophila  and  elsewhere  are  recessive, 
and  practically  all  are  in  some  way  deleterious.  Even  if  dominant  and  re- 
cessive mutations  were  occurring  with  equal  frequency,  the  deleterious  mu- 
tations in  a  population  at  any  time  would  be  mostly  recessive,  since  the  domi- 
nants would  be  rapidly  eliminated.  It  is  to  be  expected — and  it  has  been  often 
observed — that  at  most  unfixed  loci  the  recessive  is  deleterious  in  compari- 
son with  its  dominant  allele.^ 

Almost  thirty  years  ago  Sewall  Wright  (1922c)  wrote: 

Given  the  Mendelian  mechanism  of  heredity,  and  this  more  or  less  perfect  correlation  be- 
tween recessiveness  and  detrimental  effect,  and  all  the  long-known  effects  of  inbreeding — - 
the  frequent  appearance  of  abnormalities,  the  usual  deterioration  in  size,  fertility,  and  con- 
stitutional vigor  in  the  early  generations,  the  absence  of  such  decline  in  any  one  or  all  of 
these  respects  in  particular  cases,  and  the  fixation  of  type  and  prepotency  attained  in  later 
generations — are  the  consequences  to  be  expected. 

It  has  been  shown  many  times  that  populations  actually  contain  a  large 
number  of  detrimental  recessives — sufficient  to  account  for  a  large  decline  in 
vigor  on  inbreeding.  In  Drosophila  pseudoobscttra,  Dobzhansky  et  al.  (1942) 
found  that  almost  every  fly  examined  had  at  least  one  concealed  lethal.  Fur- 
ther evidence  that  at  least  some  heterosis  is  due  to  dominant  favorable  genes 
is  provided  by  the  experiments  of  Richey  and  Sprague  (1931)  on  convergent 
improvement  in  corn. 

2.  I  consider  the  statement  that  a  dominant  is  beneficial  and  the  statement  that  a  reces- 
sive is  deleterious  as  meaning  the  same  thing.  Since  a  geneticist  ordinarily  can  study  gene 
effects  only  by  substituting  one  allele  for  the  other,  he  cannot  distinguish  what  each  factor 
is  doing  individually  or  whether  it  is  harmful  or  beneficial  except  relative  to  its  allele.  That 
is,  he  can  only  tell  what  the  effect  of  the  substitution  is. 


286  JAMES  F.  CROW 

OVERDOMINANCE 

The  concept  of  a  stimulating  effect  of  hybridization  began  independently 
with  Shull  (1908, 1911b)  and  East  (1908).  It  was  assumed  that  there  was  a 
physiological  stimulus  to  development  which  increased  with  the  diversity 
of  the  uniting  gametes — with  increasing  heterozygosis.  East  (1936)  elabo- 
rated the  idea  further  by  postulating  a  series  of  alleles  each  having  positive 
action  functions,  and  with  these  functions  to  some  extent  cumulative.  As  the 
alleles  became  more  and  more  divergent  in  function,  the  action  was  postu- 
lated to  become  more  nearly  additive  in  the  heterozygote. 

At  the  time  when  East  and  Shull  first  formulated  the  hypothesis,  there  was 
no  direct  evidence  of  any  locus  at  which  the  heterozygote  exceeded  either 
homozygote.  For  a  number  of  years,  overdominance  as  an  explanation  of 
heterosis  largely  was  given  up  because  of  the  failure  to  find  such  loci. 

Stadler  (1939)  pointed  out  that  in  certain  of  the  R  alleles  in  corn  a  situa- 
tion obtains  in  which  certain  heterozygotes  have  more  areas  pigmented  than 
either  homozygote.  He  suggested  that  genes  acting  in  this  manner  could  re- 
sult in  overdominance  for  such  characters  as  size  and  yield.  Other  such  loci 
are  known  in  corn. 

There  are  now  several  cases  in  the  literature  of  single  genes  with  heterotic 
effects.  In  most  of  these  it  is  not  possible  to  rule  out  the  possibility  of  close 
linkages  giving  pseudo-overdominant  effects.  In  particular,  many  cases  may 
turn  out  to  be  pseudoallelism,  but  the  consequences  for  the  animal  or  plant 
breeder  would  not  be  changed. 

Several  workers  (Teissier,  1942a;  Robertson,  unpublished)  have  found  per- 
sistent lethals  in  Drosophila  population  cage  experiments.  If  these  are  not 
due  to  individually  heterotic  loci,  extremely  close  linkage  must  be  postu- 
lated. Also  certain  recessive  genes,  such  as  ebony,  come  to  an  equilibrium 
with  their  normal  alleles  in  population  cages.  One  of  the  most  convincing 
cases  is  that  of  the  eye  color  mutant  described  by  Buzzati-Traverso  in  this 
volume.  This  mutant  persists  in  the  population,  and  was  found  in  three  in- 
dependent stocks.  It  is  quite  improbable  that  in  each  of  these  cases  the  gene 
happened  to  be  linked  in  the  repulsion  phase  with  another  harmful  recessive. 

The  idea  of  superior  heterozygotes  has  been  upheld  by  Hull  (1945)  who 
suggested  the  word  overdominance.  Hull's  original  argument  for  overdomi- 
nance is  a  simple  one.  He  noted  that  in  most  cases  the  hybrid  between  two 
inbred  maize  lines  has  a  greater  yield  than  the  sum  of  the  two  inbreds.  This 
would  not  be  possible  with  dominant  genes  acting  in  a  completely  additive 
manner — unless  it  were  assumed  that  a  plant  with  no  favorable  dominants 
had  a  negative  yield. 

The  validity  of  this  argument  depends  on  the  unimportance  of  epistasis 
in  corn  yields.  Evidence  on  this  point  is  very  incomplete  and  somewhat  con- 
tradictory. Neal  (1935)  reported  that  the  Fo  yields  were  almost  exactly  inter- 
mediate between  the  Fi  and  the  average  of  the  parents.  This  would  suggest  tha  t 


DOMINANCE  AND  OVERDOMINANCE  287 

epistatis  is  not  important  or  else  that  there  is  some  sort  of  cancelling  out  of 
various  effects.  On  the  other  hand,  Stringfield  (1950)  found  that  in  many 
cases  backcrosses  showed  consistently  higher  yields  than  the  F2.  This  sug- 
gests some  sort  of  interaction,  as  if  some  of  the  gene  combinations  selected  for 
during  the  inbreeding  process  were  active  in  the  backcross,  but  were  broken 
in  the  F2.  None  of  these  data  give  any  evidence  as  to  the  importance  of 
epistasis  in  determining  the  difference  between  an  inbred  line  and  a  hypothet- 
ical line  with  none  of  the  favorable  dominants,  since  the  data  do  not  extend 
into  this  range.  It  is  in  this  range  where  non-additivity  might  be  expected 
to  be  most  pronounced. 

Hull's  second  argument  is  based  on  results  obtained  by  the  technique  of 
constant  parent  regression.  The  regression  of  Fi  on  one  parent,  with  the  other 
parent  held  constant,  has  different  expectations  when  there  is  overdominance 
than  when  there  is  dominance.  With  overdominance  the  regression  may  be 
negative  when  the  constant  parent  is  high-yielding,  so  the  regression  surface 
is  different  from  that  expected  with  dominance.  In  this  volume  Hull  gives 
data  which  conform  with  this  expectation. 

Overdominance  is  not  the  only  possible  explanation  of  such  results,  as 
Hull  has  pointed  out.  In  addition,  the  constant  parent  regression  technique, 
or  any  technique  making  use  of  yield  data  on  inbred  lines,  is  complicated  by 
the  difficulty  of  obtaining  consistent  results  with  inbreds.  Another  possi- 
bility is  that  the  factors  responsible  for  yield  in  inbreds  are  largely  different 
genes  from  those  determining  the  yield  in  the  hybrids.  This  possibility  will 
be  considered  later. 

For  these  reasons  it  is  still  not  possible  to  be  sure  of  the  importance  of 
overdominance  from  Hull's  methods.  They  are  at  least  strongly  suggestive, 
and  recent  data  from  Robinson  et  al.  (1949),  obtained  by  an  entirely  differ- 
ent procedure,  also  gave  evidence  of  overdominance. 

MAXIMUM  HETEROSIS  WITH  THE  DOMINANCE  HYPOTHESIS 

In  this  discussion  several  assumptions  are  made.  Most  of  these  have  been 
implicit  in  most  discussions  of  heterosis,  but  it  is  best  that  they  be  clearly  set 
forth  at  the  outset.  The  assumptions  are: 

1.  Genes  concerned  with  vigor  are  dominant,  and  in  each  case  the  domi- 
nant allele  is  beneficial  and  the  recessive  deleterious.  This  is  an  assumption 
of  convenience  which  does  not  alter  the  essential  nature  of  the  hypothesis. 
The  conclusions  still  hold  if  dominance  is  not  complete.  Also  there  are  loci 
in  which  the  recessive  is  advantageous  or  in  which  the  heterozygote  is  inter- 
mediate; but  these  are  of  no  consequence  for  heterosis  and  therefore  can  be 
omitted  from  the  discussion. 

2.  There  is  complete  additivity  of  effects  between  loci — no  epistasis. 

3.  There  are  no  barriers  to  recombination  that  prevent  each  gene  from 
reaching  its  own  equilibrium  frequency  independently  of  other  loci. 


288  JAMES  F.  CROW 

4.  The  gene  and  phenotype  frequencies  of  the  parent  population  are  at 
their  equilibrium  values. 

5.  Increased  vigor  results  in,  and  can  be  measured  in  terms  of,  increased 
selective  advantage,  though  the  selection  may  be  natural  or  artificial.  This 
assumption  restricts  the  discussion  to  those  cases  in  which  heterosis  results 
in  changes  in  the  same  direction  as  selection  had  previously  been  acting. 
Such  an  assumption  appears  to  be  valid  for  yield  characters  in  field  crops, 
and  for  viability  and  fertility  as  is  measured  in  Drosophila  population 
studies.  It  is  highly  questionable  for  such  things  as  increase  in  size  of  hybrids 
between  wild  varieties  or  species,  where  natural  selection  pressure  may  well 
have  been  toward  an  intermediate  size. 

Under  this  assumption  the  increase  of  vigor  on  hybridization  depends  di- 
rectly on  the  number  of  loci  which  are  homozygous  recessives  in  the  parent, 
but  which  become  heterozygous  in  the  hybrid.  The  individual  or  population 
of  maximum  vigor  is  one  in  which  every  allelic  pair  contains  at  least  one  domi- 
nant. The  actual  attainable  heterosis  would  be  less  than  this  in  any  particu- 
lar case. 

Consider  the  case  of  complete  dominance.  The  recessive  phenotype  is  as- 
sumed to  have  a  selective  disadvantage  of  s.  That  is,  the  dominant  and  re- 
cessive phenotypes  are  surviving  and  reproducing  in  the  ratio  of  1  to  1  —  5. 
The  rate  of  mutation  from  A  to  a  is  u  per  gene  per  generation.  Reverse  muta- 
tion will  be  ignored  as  it  can  be  shown  to  have  a  negligible  effect  on  the 
equilibrium  gene  frequency  attained. 

Genotype                 A  A  A  a  a  a 

Frequency                 P  2Q  R 

Selective  value           1  1  1  —  5 

F-\-2Q-\-R=  I 

Under  these  assumptions,  the  frequency  of  gene  A  will  be  P  -j-  ^,  while 
the  frequency  of  a  will  he  Q  -\-  R.  With  mutation  from  A  to  a  at  rate  u,  the 
frequency  of  .4  will  be  reduced  in  one  generation  by  u{P  -f-  Q)  and  the 
frequency  of  a  increased  by  the  same  amount.  Likewise,  due  to  the  effect 
of  selection,  the  frequency  of  a  will  be  decreased  by  sR.  Therefore  the  gene 
ratio,  {P  -j-  Q)/(Q  +  R),  will  change  in  one  generation  due  to  the  effects 
of  mutation  and  selection  to 

iP  +  Q){l-u) 

{P-hQ)u-\-Q-\-R-  sR' 

When  equilibrium  is  reached  the  gene  frequency  will  no  longer  change  from 
generation  to  generation  which,  stated  algebraically,  is 

P^^  ^         {P  +  Q){l-u) 

Q^R      {P-\-Q)u-\-Q-\-R-  sR- 


DOMINANCE  AND  OVERDOMINANCE  289 

This  has  the  sokition,  R  =  u/s.  (For  a  more  pedantic  demonstration  of  this, 
see  Crow,  1948.) 

The  average  reduction  in  selective  value  of  the  population  due  to  a  detri- 
mental factor  will  be  the  product  of  the  selective  disadvantage  of  the  factor 
and  the  proportion  of  individuals  possessing  the  factor.  This  amounts  to 
(s)  (ti/s),  or,  simply,  u,  the  mutation  rate.  Hence,  the  effect  of  a  detrimental 
gene  on  the  selective  value  of  the  population  is  equal  to  the  mutation  rate  to 
that  gene,  and  is  independent  of  the  selective  disadvantage  which  that  factor 
causes,  as  was  first  pointed  out  by  Haldane  (1937).  This  fact,  which  at  first 
appears  paradoxical,  is  readily  understandable  when  one  notes  that  a  mildly 
deleterious  mutant  persists  much  longer  in  the  population,  and  hence  affects 
many  more  individuals  than  one  which  has  a  greater  harmful  effect. 

The  total  effect  on  the  population  of  all  the  loci  capable  of  mutating  to 
deleterious  recessives  is  simply  the  sum  of  the  individual  mutation  rates  as 
long  as  the  gene  effects  are  additive.  If  there  are  n  such  loci  with  an  average 
mutation  rate  of  u,  the  net  reduction  in  selective  value  due  to  all  homozygous 
detrimental  recessives  at  all  loci  in  which  they  occur  is  nil.  This  is  also  ap- 
proximately correct  if  the  factors  are  multiplicative,  provided  the  individual 
effects  are  small. 

The  product  ml  is  probably  in  the  vicinity  of  .05  (Crow,  1948).  This  means 
that  if  all  the  deleterious  recessives  were  replaced  by  their  dominant  alleles, 
the  selective  advantage  of  an  equilibrium  population  would  be  increased  by 
about  this  amount.  This  could  be  considered  as  the  maximum  average  im- 
provement in  vigor,  as  measured  in  terms  of  selective  advantage,  that  could 
occur  due  to  hybridization.  This  means  that  the  dominance  hypothesis  can- 
not, under  the  conditions  postulated,  account  for  average  increases  of  more 
than  a  few  per  cent  in  vigor. 

There  are  several  reasons  why  the  5  per  cent  figure  given  above  may  be 
too  large.  One  is  that  many  deleterious  factors  considered  to  be  recessive 
may  not  be  completely  recessive.  Stern  and  Novitski  (1948)  and  Muller 
(1950)  have  shown  that  the  majority  of  lethals  and  detrimentals  that  occur 
in  laboratory  cultures  of  Drosophila  are  not  completely  recessive.  Even  if  the 
detrimental  effect  of  the  heterozygote  is  much  less  than  that  of  the  homozy- 
gote,  the  greatest  selection  effect  will  still  be  on  heterozygotes  because  of 
their  much  greater  frequency  in  the  population.  Thus,  from  the  population 
standpoint,  these  factors  would  be  acting  more  like  dominants  than  reces- 
sives. This  means  that  each  locus  would  have  a  detrimental  effect  of  2u  in- 
stead of  u  (since  a  dominant  gene  would  be  responsible  for  twice  as  many 
"genetic  deaths"  as  a  recessive),  but  the  locus  would  be  unimportant  for 
heterosis.  Since  the  n  in  the  formula  refers  only  to  the  number  of  loci  which 
are  capable  of  mutating  to  a  completely  recessive  allele,  its  value  may  be 
smaller  than  previously  assumed  and  the  product  nu  proportionately  less. 

It  has  been  assumed  that  the  parent  populations  are  at  equilibrium  be- 


290  JAMES  F.  CROW 

tween  selection  and  mutation  pressures.  This  assumption  probably  is  not 
strictly  correct  for  any  population.  Any  equilibrium  involving  occurrences 
as  rare  as  mutations  must  be  slow  of  attainment.  Hence  many  if  not  most 
populations  must  not  be  at  equilibrium.  Probably  the  most  common  way 
in  which  a  population  gets  out  of  equilibrium  is  by  an  alteration  of  the  breed- 
ing structure  or  population  number  so  that  the  effective  amount  of  inbreeding 
is  changed.  If  the  change  in  population  structure  is  such  as  to  increase  the 
amount  of  homozygosity,  a  new  equilibrium  is  reached  comparatively  rapidly 
through  the  elimination  by  selection  of  the  recessives  which  have  been  made 
homozygous.  On  the  other  hand,  if  the  change  in  population  is  such  as  to 
decrease  the  amount  of  homozygosity  a  new  equilibrium  is  attained  only 
through  the  accumulation  of  new  mutations.  This  is  an  extremely  slow 
process. 

Since  the  return  to  equilibrium  is  much  slower  when  the  population 
changes  in  the  direction  of  less  inbreeding,  it  follows  that  most  populations 
which  are  out  of  equilibrium  will  be  out  in  the  direction  of  having  too  few 
detrimental  recessives.  Therefore  the  effect  of  fluctuations  in  population 
size  and  breeding  structure  will  be  on  the  average  such  as  to  increase  the 
fitness  of  the  population.  For  this  reason,  the  average  loss  of  fitness  per  locus 
is  probably  less  than  the  mutation  rate.  Fisher  (1949)  has  pointed  out  that 
if  the  yield  of  a  crop  is  near  a  "ceiling,"  the  relative  effect  of  each  factor  con- 
ditioning yield  becomes  less.  There  will  be  a  similar  tendency  for  the  popula- 
tion to  be  out  of  equilibrium  because  of  the  slowness  of  occurrence  of  the 
mutations  required  to  bring  the  population  to  the  new  equilibrium  level. 

Another  factor  also  pointed  out  by  Fisher  is  that  complete  lethals  and 
highly  deleterious  factors  contribute  to  the  mutation  rate  but,  at  least  in 
grain  crops,  have  no  appreciable  effect  on  yield  since  they  are  crowded  out 
by  other  plants. 

All  of  these  factors  make  the  5  per  cent  figure  an  overestimate,  so  it  should 
be  regarded  as  a  maximum.  The  true  value  may  be  much  less.  In  this  con- 
nection Fisher  (1949)  said: 

...  it  would  appear  that  the  total  elimination  of  deleterious  recessives  would  make  less 
difference  to  the  yield  of  cross-bred  commercial  crops  than  the  total  mutation  rate  would 
suggest.  Perhaps  no  more  than  a  1  per  cent  improvement  could  be  looked  for  from  this 
cause.  Differences  of  the  order  of  20  per  cent  remain  to  be  explained. 

These  considerations  make  it  difficult  to  explain,  in  terms  of  the  domi- 
nance hypothesis,  cases  in  which  two  equilibrium  populations  produce  hy- 
brids with  considerable  heterosis,  or  in  which  crosses  between  inbred  lines 
average  appreciably  more  than  the  randomly  mating  populations  from 
which  they  were  derived. 

This  discussion  is  relevant  only  when  the  character  is  measurable  in  terms 
of  selective  value.  For  yield  characters  subject  to  any  high  degree  of  artificial 
selection  an  increase  in  yield  is  probably  accompanied  by  a  greater  proper- 


DOMINANCE  AND  OVERDOMINANCE  291 

tional  increase  in  selective  value.  Thus  any  conclusions  about  maximum  i)ro- 
portional  increase  in  selective  value  would  hold  a  forliori  for  yield.  Fisher 
(1949)  reaches  a  similar  conclusion  when  he  says:  "If  the  chance  of  survival  is 
equated  to  the  yield,  as  is  reasonable  with  grain  crops." 

Another  assumption  is  that  the  hybrids  are  compared  with  equilibrium 
populations.  There  is  room  for  question,  particularly  with  domestic  plants 
and  animals,  as  to  whether  selection  has  been  occurring  long  enough  and  its 
direction  has  been  consistent  enough  for  a  gene  frequency  equilibrium  to  have 
been  attained.  Another  point  that  must  be  remembered  in  discussions  of 
maize  is  that  commercial  hybrids  are  not  random  combinations  of  inbred 
lines,  but  highly  select  combinations.  An  average  hybrid  may  have  a  yield 
very  close  to  that  of  a  randomly  mating  population.  Thus  the  argument  of 
this  section  may  not  be  relevant  for  corn.  But  it  can  hardly  be  true  that  the 
high  yield  of  certain  corn  hybrids  is  due  to  the  elimination  of  deleterious 
recessives  during  inbreeding. 

The  quantitative  limit  placed  on  average  improvement  on  hybridization 
with  the  dominance  hypothesis  does  not  hold  for  overdominant  loci.  A  locus 
at  which  the  homozygote  A  A  has  a  selective  disadvantage  of  .y  with  respect 
to  the  heterozygote,  and  the  homozygote  A' A'  has  a  disadvantage  of  /,  will 
come  to  equilibrium  with  gene  frequency  of  A  equal  to  t/(s  +  0,  and  the 
frequency  of  A'  equal  to  s/(s  +  /)  (Wright,  1931b;  Crow,  1948).  The  average 
reduction  in  selective  advantage  of  the  population  due  to  the  two  homozy- 
gous genotypes  comes  out  to  be  st/{s  +  /).  The  loss  in  fitness  of  the  popula- 
tion is  of  the  order  of  magnitude  of  the  selection  coefficients,  as  Haldane 
(1937)  has  first  shown,  whereas  with  a  detrimental  recessive,  the  loss  is  of 
the  order  of  the  mutation  rate.  Hence  a  single  overdominant  locus  has  a 
tremendously  greater  effect  on  the  population  fitness  than  a  single  locus  with 
dominance  or  intermediate  heterozygote.  If  such  loci  are  at  all  frequent  they 
must  be  important.  The  question  is:  how  frequent  are  they? 

Even  with  overdominance  it  is  difficult  to  understand  large  average  in- 
creases in  selective  advantage  of  hybrids  between  equilibrium  populations. 
Such  populations  should  be  somewhere  near  their  optimum  gene  frequencies, 
which  means  that  the  hybrids  would  be  about  the  same  as  the  parents.  It 
may  be  that,  on  the  average,  hybrids  do  not  greatly  exceed  their  parents  in 
selective  advantage,  and  that  the  cases  of  increased  size  observed  in  variety 
crosses  and  occasionally  in  species  crosses  are  nothing  but  luxuriance.  If  so, 
they  are  much  less  difficult  to  explain. 

As  Bruce  showed  in  1910,  if  the  parents  differ  at  all  in  gene  frequencies, 
the  hybrids  will  be  more  heterozygous.  If  both  parents  are  at  equilibrium 
they  should  have,  for  additive  genes,  approximately  the  same  frequencies. 
But  what  differences  there  are — due  to  chance,  for  example — will  amount  to 
much  more  in  an  overdominant  than  in  a  dominant  locus  because  the  former 
has  a  gene  frequency  much  nearer  .5. 


292  JAMES  F.  CROW 

POPULATION  VARIANCE 

The  same  considerations  which  show  that  an  overdominant  locus  has  a 
much  greater  effect  on  average  population  fitness  than  a  dominant  locus  also 
show  that  an  overdominant  locus  has  a  much  greater  effect  on  the  population 
variance.  If  the  selective  values  of  the  three  genotypes,  AA,Aa,  and  aa  are  1, 
1,  and  1  —  s  respectively,  the  frequency  of  aa  genotypes  is  u/s  and  the  aver- 
age selective  value  1  —  u.  The  variance  in  fitness  will  be  sii.  On  the  other 
hand,  with  an  overdominant  locus  where  the  fitnesses  of  the  three  genotypes 
are  1  —  ^,  1,  and  1  —  ^,  the  mean  fitness  is  1  —  s/2.  The  variance  in  fit- 
ness is  5-/4. 

The  ratio  of  these  variances  is  s/4u,  which  means  that  an  overdominant 
locus  causes  a  population  variance  s/4:U  times  as  great  as  that  resulting  from 
a  recessive  locus  of  the  same  selective  disadvantage.  If  4u  is  10"^,  this 
amounts  to  100  for  ^  =  .001,  or  is  1000  for  s  =  .01.  This  makes  an  over- 
dominant  locus  with  these  selective  values  equivalent  to  100  or  1000  ordinary 
loci  in  its  effect  on  the  population  variance.  Haldane  (1950)  has  emphasized 
the  importance  of  loci  with  adaptively  superior  heterozygotes  in  increasing 
the  variance  of  natural  populations. 

From  this  we  must  conclude  that  there  doesn't  have  to  be  a  very  high 
proportion  of  overdominant  loci  for  overdominance  to  be  the  most  important 
factor  in  the  genetic  variance  of  the  population.  If  much  of  the  genetic  vari- 
ance of  a  population  is  due  to  overdominance,  this  would  explain  the  great 
slowness  of  selection.  Characters  with  high  genetic  determination  but  low 
parent-offspring  correlation  might  be  due  to  this  cause. 

The  facts  of  hybrid  corn  also  are  consistent  with  this.  Ordinary  selection 
has  not  been  effective.  Yet  there  is  a  great  deal  of  variation  in  an  open- 
pollinated  variety.  It  has  been  relatively  easy  to  find  combinations  of  inbred 
lines  that  have  yields  well  above  the  open-pollinated  averages.  There  appears 
to  be  a  relatively  high  degree  of  genetic  determination  of  yield,  but  relatively 
low  heritability.  These  results  are  not  impossible  with  dominant  genes,  es- 
pecially with  epistasis,  but  are  precisely  what  would  be  expected  if  some  of 
the  variance  were  due  to  overdominant  loci. 

A  population  with  many  overdominant  loci  is  always  well  below  its  maxi- 
mum possible  fitness.  It  is  expected  that  such  factors  could  eventually  be 
replaced  in  long  evolutionary  periods.  This  might  occur  by  an  appropriate 
mutation,  by  duplication,  or  by  modifiers.  Or  a  population  with  too  many 
overdominant  loci  might  disappear  due  to  inter-population  competition.  But 
at  any  particular  time,  a  population  may  have  a  small  proportion  of  such  loci, 
and  it  does  not  require  many  for  these  to  be  the  major  source  of  variation. 

DO  THE  SAME  GENES  DETERMINE  VARIATION  IN 
INBREDS  AND  HYBRIDS? 

The  rarer  a  recessive  phenotype  is  in  a  population,  the  greater  will  be  its 
relative  increase  in  frequency  on  inbreeding.  If  the  frequency  of  the  recessive 


DOMINANCE  AND  OVERDOMINANCE  293 

gene  is  q,  the  frequency  of  recessive  homozygotes  in  a  randomly  mating 
population  is  (f.  With  increasing  amounts  of  inbreeding,  the  frequency 
changes  from  q-  to  q.  The  smaller  the  value  of  q,  the  greater  is  the  ratio  of 
q  to  q~.  If  a  gene  is  highly  deleterious  it  will  be  very  rare  in  the  population. 
Hence  the  genotypes  which  are  most  deleterious  are  those  which  have  the 
greatest  relative  increase  in  frequency  on  inbreeding. 

These  relationships  are  brought  out  in  the  following  figures,  based  on  a 
mutation  rate  of  10~^.  The  ratio  given  is  the  ratio  of  homozygous  recessives 
in  a  homozygous  population  as  compared  with  one  which  is  mating  at 
random. 

Selective  disadvantage  (5) 0001  .001  .01  .1  lethal 

Gene  frequency  (?) 1  -032  .01  .003  .001 

Ratio  (g/5=) 10  32  100  316  1000 

This  means  that  highly  deleterious  recessives,  which  ordinarily  have  an 
effect  on  the  population  only  of  the  order  of  the  mutation  rate,  become  much 
more  important  wdth  inbreeding  and  may  become  the  major  factors  in  deter- 
mining the  fitness  of  an  inbred  population.  This  might  to  some  extent  be 
offset  by  selection  during  the  inbreeding  process,  but  such  selection  would  be 
directed  against  factors  which  are  of  no  consequence  in  a  more  heterozygous 
population. 

The  detrimental  recessive  factors  referred  to  here  include  the  lethals  and 
semilethals  (such  as  chlorophyll  deficiencies)  that  show  up  during  inbreed- 
ing. But  more  important  are  the  larger  number  of  factors,  not  individually 
detectable,  which  collectively  result  in  the  loss  of  vigor  with  inbreeding  de- 
spite rigorous  selection. 

On  the  other  hand,  the  major  part  of  the  variance  of  a  non-inbred  popu- 
lation may  well  be  determined  by  genes  of  intermediate  frequencies,  from 
.1  to  .9.  The  effect  of  such  factors  in  determining  the  population  variance 
in  fitness  would  change  only  slightly  with  inbreeding. 

As  an  example,  consider  a  hypothetical  population  mating  at  random 
whose  variance  is  made  up  of  two  components.  Ninety  per  cent  of  the  vari- 
ance is  due  to  relatively  common  loci  with  gene  frequencies  of  the  order  of  .5. 
The  other  10  per  cent  is  due  to  loci  with  recessive  gene  frequencies  of  the 
order  of  .01  or  less.  Now  when  this  population  is  inbred  without  selection, 
the  variance  due  to  the  common  genes  will  not  change  greatly  but  the  vari- 
ance due  to  the  recessive  loci  will  increase  by  a  hundred  fold  or  more.  Thus 
the  factors  which  originally  contributed  only  10  per  cent  to  the  variance 
may  now  contribute  over  90  per  cent  of  the  variance  between  the  various 
inbred  lines  derived  from  the  population. 

Gene  frequencies  of  the  order  of  .5  might  result  from  several  causes.  They 
might  be  genes  which  are  advantageous  in  one  geograj)hical  location  and 
disadvantageous  in  another  so  as  to  form  a  cline.  Or  there  might  be  seasonal 
differences  in  selective  value.  They  may  be  due  to  complex  interactions  with 
other  loci  or  be  of  extremely  small  selective  advantage  or  disadvantage.  But 


294  JAMES  F.  CROW 

another  explanation  is  selective  superiority  of  heterozygotes  (Haldane,  1950), 
at  least  for  those  factors  of  importance  in  heterosis. 

If  yield  is  determined  entirely  by  dominant  factors,  the  correlation  be- 
tween inbreds  and  their  hybrids  should  be  positive.  If  it  is  due  to  over- 
dominant  loci,  the  correlation  should  be  generally  positive,  though  there 
would  be  negative  correlations  between  yield  of  hybrids  and  inbreds  when 
the  other  inbred  is  constant  and  high  yielding.  If  both  factors  are  involved 
and  overdominant  loci  are  relatively  important  in  hybrids  while  dominants 
are  important  in  inbreds,  the  correlation  would  approach  zero.  The  experi- 
ence of  corn  breeders  has  been  that  selection  for  yield  during  inbreeding  is 
relatively  ineffective,  and  that  the  correlation  of  hybrid  with  inbred  yield, 
though  positive,  is  small. 

With  overdominant  loci  the  effect  of  a  certain  percentage  increase  in 
heterozygosity  is  to  cause  the  vigor  to  increase  by  a  certain  amount.  De- 
creasing the  heterozygosity  by  the  same  percentage  would  cause  a  decrease 
of  approximately  the  same  amount.  On  the  other  hand,  with  dominant  loci, 
making  the  original  equilibrium  population  more  heterozygous  would  cause 
a  very  slight  increase,  whereas  making  the  population  more  homozygous 
would  have  a  decreasing  effect  of  a  much  greater  amount.  Therefore  it  is 
easier  to  account  for  inbreeding  depression  by  dominant  loci  than  to  account 
for  increase  in  vigor  on  hybridization  above  the  level  of  a  random  mating 
population. 

I  should  like  to  suggest  the  following  interpretation  of  the  effects  of  in- 
breeding and  hybridization:  The  deleterious  effects  of  inbreeding  and  the  re- 
covery on  hybridization  are  mainly  due  to  loci  where  the  dominant  is  favor- 
able and  the  recessive  allele  so  rare  as  to  be  of  negligible  importance  in  a  non- 
inbred  population.  Variance  of  a  non-inbred  population,  and  hybrid  vigor 
when  measured  as  an  increase  over  an  equilibrium  population,  are  deter- 
mined largely  by  genes  of  intermediate  frequency,  probably  mostly  over- 
dominants. 

OVERDOMINANCE  AND  GENE  ACTION 

In  order  to  have  overdominance  it  is  not  necessary  that  the  immediate 
gene  products  of  the  heterozygote  exceed  in  quantity  or  variety  those  of 
either  homozygote.  At  the  level  of  the  immediate  gene  product,  or  any  inter- 
mediate state,  the  effect  of  the  heterozygote  may  be  intermediate  between 
the  two  homozygotes  and  still  result  in  a  greater  final  result.  Any  kind  of 
situation  in  which  something  is  produced  for  which  an  intermediate  amount 
is  optimum  could  be  such  that  the  heterozygote  is  nearer  this  optimum  than 
either  homozygote. 

A  model  for  such  cases  is  found  in  the  sulfanilamide-requiring  strain  of 
Neurospora  reported  by  Emerson  (1948).  When  this  mutant  is  present  the 
heterokaryotic  state  of  the  suppressor  gene  results  in  more  nearly  the  opti- 


DOMINANCE  AND  OVERDOMINANCE  295 

mum  amount  of  para-amino  benzoic  acid  than  either  homokaryon.  Other 
cases,  less  known  biochemically,  may  be  similar. 

I  think  that  it  is  doubtful  whether  such  a  system  would  persist  for  long 
evolutionary  periods.  Alleles  of  intermediate  productivity  could  arise  and 
replace  the  originals.  Also  modiiiers  altering  the  expression  of  the  homozy- 
gotes  would  have  considerable  selection  pressure.  Or  if  the  alleles  were  anti- 
morphic,  the  situation  might  be  resolved  by  duplication,  as  Haldane  (1937) 
has  suggested.  It  is  significant  that  the  system  reported  by  Emerson  is  not 
one  which  is  ordinarily  of  importance,  but  acts  only  in  the  presence  of  the 
sulfanilamide-requiring  mutant. 

A  form  of  gene  action  that  appears  more  likely  to  account  for  instances 
of  overdominance  is  one  in  which  the  two  alleles  differ  qualitatively  or  each 
does  something  that  the  other  fails  to  do.  Instances  of  mosaic  dominance 
provide  excellent  examples.  This  has  been  demonstrated  for  the  scute  series 
of  bristle  characters  in  Drosophila  and  for  color  pattern  in  beetles  (Tan, 
1946).  Other  examples  are  provided  by  the  ^4  and  R  loci  in  maize. 

Similar  examples  of  physiological  mosaic  dominance  are  found  where  the 
heterozygote  apparently  produces  something  approximating — at  least 
qualitatively — the  total  effect  of  the  two  homozygotes.  An  example  is  rust 
resistance  in  flax,  where  each  strain  is  resistant  to  a  certain  rust  but  the  hy- 
brid is  resistant  to  both  (Flor,  1947).  By  the  usual  tests  for  allelism,  the  two 
resistance  factors  are  alleles.  Another  series  of  examples  is  found  in  the 
blood  group  antigens  in  man,  cattle,  and  elsewhere.  In  almost  every  instance 
the  heterozygote  has  all  the  antigenic  properties  of  both  homozygotes 
(Irwin,  1947).  The  presence  of  both  the  normal  and  abnormal  types  of 
hemoglobin  in  humans  heterozygous  for  the  gene  for  sicklemia  provides 
another  example  (Pauling,  1950). 

Many  instances  of  overdominance  may  have  a  similar  explanation.  This 
is  the  kind  of  action  that  East  (1936)  postulated  in  his  discussion  of  heterosis 
due  to  cumulative  action  of  divergent  alleles.  It  is  not  necessary  that  the 
effects  be  completely  cumulative;  only  that  the  net  effect  on  the  phenotype 
be  greater  in  the  heterozygote  than  in  the  homozygote.  Any  system  in  which 
the  alleles  act  on  different  substrates  to  produce  the  same  or  different  prod- 
ucts, or  convert  the  same  substrate  into  different  products — neomorphs,  in 
Muller's  terminology — could  result  in  overdominance. 

Any  of  the  examples  listed  above  may  turn  out  to  be  closely  linked  genes 
(pseudoalleles)  rather  than  alleles.  In  most  cases  it  is  impossible  to  distin- 
guish between  these  alternatives.  If  the  overdominance  effect  is  due  to 
linked  genes,  eventually  a  crossover  should  result  in  a  situation  where  the 
desirable  effects  could  be  obtained  in  a  homozygous  individual.  If  there  are 
position  effects,  it  may  be  that  no  homozygous  arrangement  is  as  advanta- 
geous as  one  which  is  heterozygous.  Unless  there  are  position  effects,  it  does 
not  seem  likely  that  heterosis  due  to  pseudoallelism  would  persist  for  any 


296  JAMES  F.  CROW 

great  length  of  time,  but  in  any  particular  population  such  factors  might  be 
important. 

IS  INCREASED  SIZE  ADAPTIVE? 

The  foregoing  arguments  are  based  on  the  assumption  that  heterosis  is 
measurable  in  terms  of  increased  selective  advantage.  The  selection  may  be 
natural  or  man-imposed.  This  assumption  would  appear  to  be  reasonable  for 
such  factors  as  fertility  and  resistance  to  disease.  It  also  would  apply  to  in- 
crease in  size  or  yield,  if  the  direction  of  selection  in  the  past  were  in  this  di- 
rection, as  in  corn.  However,  it  is  questionable  whether  the  increase  in  size 
that  is  sometimes  observed  in  variety  hybrids  is  really  adaptive. 

Mather  (1943)  and  especially  Dobzhansky  (1950)  have  emphasized  that 
increased  size  does  not  necessarily  result  in  increased  fitness  in  natural  popu- 
lations. Dobzhansky  proposed  the  words  euheterosis  and  luxuriance,  re- 
spectively, for  increased  selective  advantage  and  for  mere  non-adaptive  in- 
crease in  size.  In  these  terms  this  discussion  has  dealt  entirely  with  eu- 
heterosis. 

If  euheterosis  occurs  in  species  or  variety  crosses,  it  is  very  difficult  to 
explain.  It  raises  the  troublesome  question:  How  can  the  hybrid  between 
two  well  adapted  strains  be  better  adapted  than  its  parents  when  there  has 
been  no  selection  in  the  past  for  its  adaptation?  It  may  be  that  euheterosis 
is  developed  only  under  some  form  of  selection,  as  in  the  inversion  heterozy- 
gotes  studied  by  Dobzhansky,  or  in  the  series  of  hybrids  between  inbred  lines 
of  corn  selected  for  combining  ability. 

If  large  size  is  not  advantageous,  luxuriance  may  be  due  to  the  covering 
of  recessive  factors  which  were  acting  as  size  bottlenecks  and  had  been 
selected  into  the  population  because  of  this.  Each  of  the  parents  might  have 
its  growth  limited  by  or  held  in  check  by  a  series  of  factors,  and  if  some  of 
these  were  recessive,  increased  size  would  be  found  in  the  hybrids. 

SUMMARY 

Since  the  earliest  attempts  to  explain  hybrid  vigor  in  Mendelian  terms 
there  have  been  two  principal  hypotheses.  The  first  of  these  is  the  domi- 
nance hypothesis.  This  notes  the  observed  correlation  between  recessiveness 
and  detrimental  effect  and  attributes  the  increased  vigor  of  heterozygosity 
to  the  covering  of  deleterious  recessive  factors  by  their  dominant  alleles. 
The  alternative  hypothesis,  the  overdominance  hypothesis,  assumes  that 
heterozygosity  per  se  is  important — that  there  exist  loci  at  which  the  hetero- 
zygote  is  superior  to  either  homozygote. 

It  is  clear  that  the  dominance  hypothesis  is  adequate  to  explain  the  de- 
terioration that  results  from  inbreeding  and  the  recovery  of  vigor  on  out- 
crossing, but  it  is  difficult  to  explain  how  the  hybrids  could  greatly  exceed  in 
fitness  the  equilibrium  populations  from  which  their  parents  were  derived. 
The  overdominance  hypothesis  demands  the  assumption  of  a  kind  of  gene 


DOMINANCE  AND  OVERDOMINANCE  297 

action  known  to  be  rare,  but  it  is  pointed  out  that  if  only  a  small  proportion 
of  the  loci  are  of  this  type,  these  may  nevertheless  be  the  major  factor  in  the 
population  variance. 

The  following  interpretation  is  suggested:  Inbreeding  depression  and  re- 
covery on  crossing  are  mainly  the  result  of  loci  at  which  the  favorable  allele  is 
dominant  and  the  recessives  are  at  low  frequency.  On  the  other  hand  the 
variance  of  heterozygous  populations  and  the  differences  between  different 
hybrids  are  due  mainly  to  loci  with  intermediate  gene  frequencies.  It  appears 
likely  that  such  loci  are  due  to  selectively  superior  heterozygotes,  but  there 
are  several  other  possibilities. 


LEROY  POWERS 

USDA,  Bureau  of  Plant  Industry 


Chapter  19 

Gene  Recombinotion 
and  Heterosis 


This  article  will  be  confined  primarily  to  the  tomato  {Lycopersicon)  genetic 
work  which  has  a  bearing  on  gene  recombination  and  heterosis.  The  barley 
(Hordeum)  genetic  research  which  is  discussed  briefly  was  conducted  at  the 
University  of  Minnesota.  The  tomato  genetic  research  which  constitutes  the 
bulk  of  the  material  discussed  was  conducted  at  the  United  States  Horticul- 
tural Field  Station,  Cheyenne,  Wyoming. 

With  the  present  available  methods  of  analysis  it  is  difficult  in  quantita- 
tive inheritance  studies  to  distinguish  between  blocks  of  fairly  closely  linked 
genes  and  individual  pairs  of  genes.  This  has  been  shown  by  the  work  of 
Jones  (1917),  Warren  (1924),  Mather  (1942,  1949),  and  Straus  and  Gowen 
(1943).  Consequently,  in  this  article  where  the  two  genetic  systems  are  not 
distinguishable  the  term  pairs  of  genes  will  be  employed.  Mather  (1949)  has 
used  the  term  ejffedive  factor  to  depict  such  a  genetic  situation. 

MARKER  GENES  AND  LINKAGE  IN  BARLEY 

Powers  (1936)  has  shown  that  in  a  cross  between  Bl  {Hordeum  deficieiis) 
and  Brachytic  {Hordeum  vulgare)  the  Fi,  which  is  a  two-row  barley,  gave  a 
greater  yield  of  seed  per  plant  than  either  the  two-row  or  six-row  parents. 
Then,  weight  of  seed  per  plant  shows  heterosis.  The  data  on  marker  genes 
and  linkage  in  barley  presented  have  some  bearing  upon  whether  any  of  the 
advantages  of  the  Fi  hybrid  attributable  to  heterosis  can  be  recovered  in 
inbred  lines  through  gene  recombinations. 

The  deficiens  (two-row)  character  was  found  to  be  differentiated  from  the 
vulgare  (six-row)  character  by  one  pair  of  genes  designated  as  Vv,  and  the 
brachytic  character  from  the  normal  character  by  one  pair  of  genes  designated 

298 


GENE   RECOMBINATION  AND   HETEROSIS 


299 


as  Brbr.  Using  these  symbols,  the  genotype  of  the  Fi  is  VvBrbr.  The  Vv  gene 
pair  is  carried  on  chromosome  1  and  the  Brbr  gene  pair  on  chromosome  7. 

Table  19.1  gives  the  comparative  effect  upon  four  quantitative  characters 
of  genes  associated  in  inheritance  with  Vv  and  vv  and  VV  and  vv,  as  deter- 

TABLE  19.1 

COMPARATIVE   EFFECT    UPON  FOUR   QUANTITATIVE   CHARACTERS    OF 

GENES  ASSOCIATED  IN  INHERITANCE  WITH  Vv  AND  vv,  AND 

VV  AND  vv;  F-,  GENOT\TES  OF  A  BARLEY  HYBRID 


Weight 

3F  Seed* 

Spikes  per  Pl.ant* 

Height  of  Plant* 

Length 

3F  Awn* 

Genotvpe 

Vv-vv 

vv-vv 

Vv-vv 

VV-VV 

Vv-vv 

vv-vv 

Vv-vv 

VV-VV 

BrBr 

Brbr 

brbr 

-2.22 
-2.98 
-1.88 

-3.44 
-3.74 
-2.74 

1.72 
0.94 
0.13 

0.21 

0.39 

-0.94 

1.54 
2.08 
1.03 

0.64 

1.41 

-0.68 

16.58 

16.42 

1.95 

7.. SO 

9.20 

-6.68 

*  Weight  of  seed  per  plant  is  expressed  in  grams,  spikes  per  plant  in  number,  height  of  plant  in  inches,  and 
length  of  awn  in  millimeters. 


TABLE  19.2 

COMPAR.VTIVE   EFFECT   UPON  FOUR   QUANTITATIVE   CHARACTERS    OF 

GENES  ASSOCIATED  IN  INHERITANCE  WITH  Vv  AND   VV,  AND 

VV  AND  vv;  F.  GENOT\TES  OF  A  BARLEY  HYBRID 


Weight  ( 

3F  Seed* 

Spikes  per  Plant* 

Height  of  Plant* 

Length 

OF  Awn* 

Genotype 

Vv-vv 

VV-VV 

Vv-vv 

vv-vv 

Vv-vv 

vv-vv 

Vv-vv 

VV-VV 

BrBr 

Brbr 

brbr 

1.22 
0.76 
0.86 

-3.44 
-3.74 
-2.74 

1.51 
0.55 
1.07 

0.21 

0.39 

-0.94 

0.90 
0.67 
1.71 

0.64 

1.41 

-0.68 

9.08 
7.22 
8.63 

7.50 

9.20 

-6.68 

*  Weight  of  seed  per  plant  is  expressed  in  grams,  spikes  per  plant  in  number,  height  of  plant  in  inches,  and 
length  of  awn  in  millimeters. 


mined  by  differences  between  means  of  F2  plants.  In  every  case,  the  differ- 
ences between  Vv  and  vv  are  greater  than  the  differences  between  VV  and  vv 
for  spikes  per  plant,  height  of  plant,  and  length  of  awn.  With  the  exception 
of  the  comparison  between  VV  and  vv  within  the  brbr  genotype,  the  differ- 
ences are  in  favor  of  the  two-row  (Vv  and  VV)  segregates  as  compared  with 
the  six-row  (vv)  segregates.  Within  the  brbr  genotype,  iJt)  plants  exceed  the  VV 
plants  for  all  three  characters.  As  regards  weight  of  seed  per  plant  in  every 
case  the  six-row  plants  outyielded  the  two-row  plants  whether  heterozygous 
deficiens  or  homozygous  deficiens.  However,  the  differences  between  vv  and 
Vv  were  less  than  those  between  vv  and  VV. 

The  data  of  Table  19.2  show  that  for  all  characters  the  Vv  plants  give  an 


300  LEROY  POWERS 

increase  over  the  VV  plants,  and  with  the  exception  of  the  Brbr  genotype  for 
height  of  plant  and  length  of  awn,  the  differences  of  Vv-VV  are  greater  than 
the  differences  for  VV-vv. 

These  facts  concerning  the  data  reveal  that  Vv  is  associated  with  an  in- 
crease in  all  four  quantitative  characters.  For  spikes  per  plant,  height  of 
plant,  and  length  of  awn  this  increase  results  in  heterosis. 

Hypotheses  for  Difference  in  Vigor 

If  the  increase  noted  is  due  solely  to  an  interaction  between  V  and  v  such 
as  is  depicted  by  East's  physiological  hypothesis,  then  it  would  not  be  pos- 
sible to  obtain  homozygous  lines  possessing  any  of  this  increase.  However,  if 
the  heterosis  noted  is  due  to  a  combination  of  favorable  and  unfavorable 
genes  linked  with  V  and  v,  it  should  be  possible  to  obtain  lines  in  which 
some  of  the  favorable  genes  are  recombined.  These  lines  should  show  some 
increase  in  the  four  quantitative  characters  studied.  In  the  event  that  linkage 
of  genes  favorable  and  unfavorable  to  an  increase  in  the  quantitative  charac- 
ters was  found  to  furnish  the  most  logical  explanation,  an  intraallelic  interac- 
tion such  as  depicted  by  East's  physiological  hypothesis  still  may  be  having 
some  influence  as  the  two  systems  are  not  mutually  exclusive. 

Tables  19.1  and  19.2  show  that  Vv  results  in  an  increase  of  all  four  charac- 
ters: weight  of  seed  per  plant,  number  of  spikes  per  plant,  height  of  plant, 
and  length  of  awn.  This  fact  is  most  simply  explained  by  assuming  the  pro- 
duction of  a  favorable  growth-promoting  substance  which  influences  all  of 
them.  Then  such  being  the  case,  on  the  basis  of  East's  (1936)  physiological 
hypothesis,  V  and  v  supplement  each  other,  resulting  in  greater  development. 
Next  consider  the  development  of  the  lateral  florets  which  determines  the 
number  of  rows  of  kernels  per  spike  (two-row  or  six-row  spikes).  The  Vv 
segregates  are  two-row  types,  whereas  the  vv  segregates  are  six-row  types. 
Hence,  as  regards  the  character  number  of  rows  of  kernels  per  spike,  the  in- 
teraction between  V  and  v  is  such  as  to  prohibit  the  development  of  the 
lateral  florets,  resulting  in  a  two-row  barley  spike  rather  than  a  six.  Summing 
up,  on  the  basis  of  the  physiological  hypothesis,  in  the  case  of  four  quantita- 
tive characters  the  interaction  between  V  and  v  is  such  as  to  stimulate  de- 
velopment. In  the  case  of  number  of  rows  of  kernels  per  spike  the  interac- 
tion is  such  as  to  prohibit  development  of  the  lateral  florets.  From  physio- 
logical genetic  considerations  such  a  pleiotropic  effect  seems  rather  im- 
probable. 

Explaining  the  heterosis  associated  with  Vv  plants  on  the  basis  of  linkage, 
a  simple  interpretation  would  be  that  the  favorable  linked  genes  and  their 
alleles  interact  according  to  Jones's  (1917)  hypothesis  to  produce  a  substance 
favorable  to  growth  processes,  resulting  in  the  heterosis  noted;  and  that  V 
is  dominant  to  v  resulting  in  Vv  (Fi)  plants  having  two-row  barley  spikes. 
This  explanation  does  not  require  the  assumption  that  V  and  v  stimulate 


GENE  RECOMBINATION  AND  HETEROSIS 


301 


growth  in  one  character  and  inhibit  it  in  another,  and  hence  is  more  in  ac- 
cord with  modern  physiological  genetic  concepts. 

The  article  by  Powers  (1936)  furnishes  additional  information  j)ertaining 
to  gene  recombination  and  heterosis.  If  genes  other  than  Vv  are  responsible 
for  the  heterosis  noted,  then  F2  plants  having  a  genotype  identical  to  the  Fi 
generation  should  give  a  somewhat  lower  yield  than  the  Fi.  Since  the  Fi 
plants  were  not  grown  in  a  randomized  experiment  with  the  F2  plants,  the 
comparison  must  be  made  through  the  Bl  parent.  As  compared  through  the 
Bl  parent  an  actual  reduction  of  one  gram  in  yield  of  seed  per  plant  was 
found  (Powers,  1936).  This  reduction  could  be  due  to  genes  carried  on  chro- 
mosome 1,  as  are  V  and  v,  or  to  genes  carried  on  other  chromosomes.  In 
either  event,  theoretically  some  of  the  genes  favorable  to  increased  weight 

TABLE  19.3 

COMPARISON  BETWEEN  PARENTS  AND  F2  PAREN- 
TAL GENOT\TES  FOR  WEIGHT  OF  SEED 
PER  PLANT  IN  A  BARLEY  HYBRID 


Total 

Weight  of  Seed  per  Plant  in  Grams 

Number  of 
Plants 

Fs 

Parent 

Differ- 
ence 

/ 

78  and  266* 

64  and  63  f  . 

3.9 
4.5 

1.9 
4.0 

2.0 
0.5 
1.5 

28.189 
0  761 

Interaction 

5  807 

*  VVBrBr,  genotype  of  Bl  parent,  two-row  normal. 

t  vvbrbr,  genotype  of  Brachytic  parent,  six-row  Brachytic. 

of  seed  per  plant  that  resulted  in  the  heterosis  noted  in  the  Fi  population 
must  be  capable  of  recombination. 

Even  though  some  of  the  genes  favorable  to  increased  growth  can  be  re- 
combined,  the  yield  of  the  lines  in  which  the  favorable  genes  have  been  com- 
bined depends  upon  the  nature  of  the  interaction  of  the  genes.  The  weights 
of  seed  per  plant  of  parents  and  F2  plants  of  the  parental  genotypes  are  given 
in  Table  19.3.  From  this  table  it  can  be  seen  that  the  F2  plants  of  the  VVBrBr 
genotype  gave  an  increased  yield  of  2.0  grams  per  plant  over  the  Bl  parental 
plants  having  the  same  genotype.  However,  the  F2  plants  of  the  same  geno- 
type as  the  Brachytic  parent  gave  an  increase  over  this  parent  of  only  0.5 
grams  per  plant,  which  is  not  statistically  significant.  The  interaction  of  1.5 
grams  (Table  19.3)  is  statistically  significant.  This  means  that  a  preponder- 
ance of  the  genes  favorable  to  increased  weight  of  seed  per  plant  must  have 
entered  the  cross  from  the  Brachytic  parent.  The  balance  of  the  unfavorable 
genes  that  entered  the  cross  from  the  Bl  parent  did  not  cause  a  correspond- 
ing decrease  in  weight  of  seed  per  plant  of  the  F2  plants  possessing  the  vvbrbr 
genotype. 


302 


LEROY  POWERS 


In  this  same  study  (Powers,  1936)  found  that  the  greater  the  number  of 
genes  in  the  genotype  tending  to  increase  a  character  the  greater  is  the  effect 
of  any  given  gene.  It  is  apparent  that  it  is  not  possible  to  definitely  predict 
the  yield  of  seed  per  plant  resulting  from  recombining  genes  favorable  to 
growth  because  of  the  interactions  noted.  Either  a  greater  or  smaller  increase 
than  expected  may  be  obtained.  Such  interactions  of  genes  would  affect  the 
yield  of  plants  in  which  the  favorable  genes  were  recombined,  and  hence  the 
feasibility  of  obtaining  inbred  lines  equaling  or  excelling  the  Fi  hybrid.  In 
some  cases  the  probability  of  getting  the  desired  results  would  be  increased 
and  in  other  cases  decreased;  depending  on  the  type  of  interallelic  and  intra- 
allelic  interactions  of  the  genes. 

GENE   RECOMBINATIONS   DIFFERENTIATING   WEIGHT   PER   LOCULE 
WHICH   EXCEED  HETEROSIS  OF  F,    POPULATION 

The  data  for  weight  per  locule  of  fruit  for  the  Porter  X  Ponderosa  to- 
mato hybrid  and  parental  populations  grown  at  Woodward,  Oklahoma,  in 

TABLE  19.4 

.ARITHMETIC  AND  LOGARITHMIC  MEANS  FOR 
WEIGHT  PER  LOCUI.E  OF  PORTER  X  PONDERO- 
SA TOMATO  HYBRID  AND  PARENTAL  POPULA- 
TIONS* 


Mea\ 

P0PUL.\TI0N 

Arithmetic 

Logarithmic 

Porter 

10.2 
11.8 
14.4 
13.5 
13.7 
9.8 

1.018253  +  0.012325 

Bi  to  Porter          .... 

1.070936  +  0.009939 

F, 

Fs 

Bi  to  Ponderosa 

Ponderosa 

1.168729  +  0.010134 
1.128481+0.011879 
1.124941+0.012651 
0.982054  +  0.011845 

*  Grown  at  Woodward,  Oklahoma,  in  1041;  original  data  taken  in  grams 
and  transformed  to  logarithms  to  obtain  the  means  and  standard  errors  of  the 
logarithms. 

1941  (Powers,  Locke,  and  Garrett,  1950)  will  be  analyzed  to  determine 
whether  in  Fo  and  backcross  populations  gene  recombinations  are  occurring 
which  exceed  the  heterosis  of  the  Fi  population. 

The  means  for  weight  per  locule  calculated  on  both  the  arithmetic  and 
logarithmic  scales  are  given  in  Table  19.4.  Weight  per  locule  is  greatest  for 
the  Fi  population,  and  the  means  of  the  Bi  to  Porter,  F2,  and  Bi  to  Ponderosa 
populations  are  larger  than  the  means  of  the  Ponderosa  and  Porter  parents, 
but  smaller  than  the  mean  of  the  Fi  population.  The  only  means  not  showing 
significant  differences  are  the  means  of  Porter  and  Ponderosa,  and  the  means 
of  the  F2  and  Bi  to  Ponderosa  populations.  Hence,  in  these  hybrid  popula- 
tions weight  per  locule  definitely  shows  heterosis  on  either  scale. 


GENE   RECOMBINATION  AND  HETEROSIS 


303 


The  frequency  distributions  for  weight  per  locule  for  the  Porter  X  Pon- 
(lerosa  hybrid  and  parental  populations  are  given  in  Table  19.5.  This  table 
shows  that  the  F2  and  Bi  to  Ponderosa  populations  have  plants  falling  into 
classes  of  greater  value  than  1.51 1883,  the  last  class  in  which  Fi  or  Ponderosa 
plants  occur.  There  are  nine  such  F2  i)lants  and  three  such  Ei  to  Ponderosa 
plants.  If  no  recombination  of  genes  to  produce  plants  with  weight  per  locule 
greater  than  the  Fi  plants  is  possible,  these  plants  with  values  greater  than 
any  individual  of  the  Fi  population  must  be  chance  deviates.  Moreover,  the 
chance  deviates  must  be  those  plants  in  the  F2  population  having  the   I'^ 

TABLE  19.5 

OBTAINED  FREQUENCY  DISTRIBUTIONS  FOR  WEIGHT  PER  LOCULE 
OF  TOMATO  FRUITS  FOR  PORTER  X  PONDEROSA  HY- 
BRID AND  PARENTAL  POPULATIONS* 


Upper  Limit  of  Class  in  Logarithms  or  Grams 

Popula- 
tion 

0 
d 

d 

00 

d 

On 
rs 

d 

00 

q 

0 

o\ 

so 
On 

q 

00 
0 

00 
•* 

00 

-0 

00 

0 

10 

00 

m 

0 
0 

00 

00 
00 

OS 

00 
IT) 

0 

00 
10 

10 

0 

0 
00 
00 

o> 

Total 
Plants 

Porter...  . 

1 

4 

27 

80 

98 

20 

2 

232 

Bi  to  Por- 

ter  

1 

3 

13 

54 

81 

102 

80 

49 

35 

16 

11 

1 

1 

1 

448 

Fi 

1 
13 

6 
31 

22 
68 

35 
8? 

49 

81 

37 
63 

34 
47 

23 
?4 

13 
17 

4 

11 

2 

8 

3 
3 

4 

1 

4 

? 

? 

1 

233 

F2 

453 

Bi  to  Pon- 

derosa. . 

1 

6 

17 

29 

45 

71 

72 

62 

52 

26 

19 

18 

1 

4 

4 

4 

1 

1 

1 

434 

Pondero- 

sa  

10 

21 

28 

25 

16 

18 

9 

4 

6 

2 

3 

1 

1 

1 

145 

*  Grown  at  Woodward,  Oklahoma,  in  1941;  original  data  taken  in  grams  and  transformed  to  logarithms  to 
obtain  the  means  and  standard  errors  of  the  logarithms. 


genotype  or  a  very  similar  genotype.  The  probability  of  their  being  chance 
deviates  possessing  the  Fi  or  similar  genotypes  can  be  determined. 

The  mean  of  the  logarithms  of  the  Fi  population  is  1.168729,  and  the 
standard  error  of  a  single  determination  is  0.123426.  Calculations  (for  meth- 
od see  Powers,  Locke,  and  Garrett,  1950)  show  that  only  0.3  per  cent  of  such 
a  genotypic  population  would  be  expected  to  have  a  value  greater  than 
1.511883.  The  following  tabulation  shows  the  theoretical  number  of  gene 
pairs  differentiating  the  parents,  the  theoretical  percentage  of  the  j)opula- 
tion  of  the  F2  or  Bi  to  Ponderosa  populations  possessing  the  same  genotyi)e 
as  the  Fi,  the  theoretical  number  of  plants  of  the  Fi  genotype  in  a  popula- 
tion of  453  F2  plants  and  in  a  population  of  434  Bi  to  Ponderosa  plants,  and 
the  theoretical  number  of  plants  of  the  Fi  genotype  in  the  F2  population 
and  in  the  Bi  to  Ponderosa  population  expected  to  exceed  a  value  of  1.511883. 

An  examination  of  the  data  opposite  one  pair  of  genes  in  the  tabulation  be- 
low shows  that  only  0.68  Fo  plants  would  be  expected  to  exceed  a  value  of 


304 


LEROY  POWERS 


1.511883,  whereas  9  plants  did  so  (see  Table  19.5).  The  same  comparison 
for  the  Bi  to  Ponderosa  population  is  0.65  expected  and  3  obtained.  Also,  a 
study  of  the  tabulation  below  reveals  that  with  an  increased  number  of 
gene  pairs  diflferentiating  the  parents  the  odds  become  even  greater  against 
those  plants  which  exceed  1.511883  being  chance  deviates. 

It  remains  to  be  seen  whether  plants  of  the  Fi  genotype  plus  plants  of 
genotypes  which  might  have  similar  effects,  but  do  not  possess  recombina- 
tion of  favorable  genes  in  excess  of  the  total  number  of  favorable  genes  car- 


Number 

Pairs  of 

Genes 

Per  Cent 
Population 
OF  Fi  Geno- 
type 

Number  of  Plants 
OF  Fi  Genotype 

Number  of  Plants 
Expected  To  Ex- 
ceed A  Value  of 
1.511883 

F2 

Bi 

F2 

Bi 

1 

50.00 

25.00 

12.50 

6.25 

3.12 

226.50 

113.25 

56.62 

28.31 

14.17 

217.00 

108.50 

54.25 

27.12 

13.56 

0.68 
0.34 
0.17 
0.08 
0.04 

0.65 

2 

3 

0.33 
0.16 

4 

0.08 

5 

0.04 

ried  by  the  Fi,  could  be  responsible  for  the  results  noted.  The  result  would  be 
to  increase  the  proportion  of  the  F2  and  Bi  to  Ponderosa  populations  fluctu- 
ating around  means  very  similar  in  magnitude  to  that  of  the  Fi  population. 
The  extreme  case  (but  highly  improbable)  would  be  to  have  all  of  these  two 
populations  made  up  of  such  plants.  On  this  basis  and  on  the  basis  that  the 
parents  are  differentiated  by  one  pair  of  genes,  the  number  of  plants  of  the 
F2  population  expected  to  exceed  1.511883  is  1.36,  and  for  the  Bi  to  Pondero- 
sa population  is  1.30.  The  number  of  plants  obtained  (Table  19.5)  is  9  for  the 
F2  population  and  3  for  the  Bi  to  Ponderosa  population.  Furthermore,  the 
Bi  to  Porter  population  had  1  plant  in  a  class  beyond  that  in  which  any  Fi 
plants  occurred. 

The  analysis  can  be  carried  further.  For  the  F2  population  the  number  of 
plants  expected  to  exceed  1.562293  is  0.3223  and  the  number  obtained  is  3. 
Whereas  the  values  for  the  Bi  to  Ponderosa  population  are  0.3087  and  1, 
respectively.  Also,  the  frequency  distributions  (Table  19.5)  in  general  do 
not  support  the  supposition  that  over  one  half  of  the  plants  of  the  F2  and  Bi 
to  Ponderosa  populations  are  fluctuating  around  a  mean  as  great  as  that  of 
the  Fi  generation.  Again  with  an  increase  in  number  of  gene  pairs  differentiat- 
ing the  parents,  the  odds  against  the  plants  exceeding  1.562293  being  chance 
deviates  become  even  greater.  It  is  evident  that  the  data  are  not  in  accord 
with  the  assumption  that  plants  of  the  Fi  genotype  have  the  greatest  weight 
per  locule.  This  is  true  regardless  of  the  number  of  gene  pairs  differentiating 
the  parents.  Therefore,  some  of  the  plants  falling  in  classes  having  values 


GENE  RECOMBINATION  AND  HETEROSIS  305 

greater  than  1.511883  must  have  genotypes  composed  of  more  favorable 
genes  than  the  Fi,  and  therefore  recombinations  of  genes  to  produce  plants 
having  a  greater  weight  per  locule  than  the  Fi  plants  have  occurred. 

Whether  inbred  lines  retaining  this  increased  weight  per  locule  can  be 
established  is  dependent  upon  the  number  of  gene  pairs  differentiating  the 
parents  and  linkage  relations  (Jones,  1917).  Close  linkage  of  genes  favorable 
to  increase  in  weight  per  locule  would  favor  recombination.  Whereas  close 
linkage  of  genes  favorable  to  increase  in  weight  per  locule  with  those  not 
favorable  would  hinder  recombination  and  hence  reduce  the  chances  of  ob- 
taining inbred  lines  retaining  some  or  all  of  the  advantages  attributable  to 
heterosis. 

The  data  furnish  evidence  concerning  the  number  of  gene  pairs  differen- 
tiating weight  per  locule.  From  Table  19.5  it  can  be  seen  that  the  plants  of 
the  F2  generation  falling  beyond  the  value  1.511883  are  distributed  over  four 
different  classes,  and  those  of  the  Bi  to  Ponderosa  population  falling  beyond 
this  same  value  occur  in  three  different  classes.  The  behavior  of  these  plants 
cannot  be  explained  on  the  basis  of  five  or  more  independently  inherited 
pairs  of  genes,  as  there  are  too  many  of  these  plants  falling  beyond  the 
1.511883  class.  In  addition,  the  weights  per  locule  of  those  falling  in  these 
classes  are  greater  than  can  be  explained  on  the  basis  of  chance  deviation. 

Further,  to  account  for  the  plants  of  the  F2  and  Bi  to  Ponderosa  popula- 
tions falling  in  those  classes  beyond  1.511883,  on  the  basis  of  five  or  more 
pairs  of  independently  inherited  genes  differentiating  the  parents,  it  would 
be  necessary  to  assume  that  50  per  cent  or  more  of  the  plants  were  fluctuating 
around  a  mean  greater  than  that  of  the  Fi  generation.  Since  the  means 
(Table  19.4)  of  the  F2  and  Bi  to  Ponderosa  populations  are  less  than  the 
mean  of  the  Fi,  these  populations  cannot  have  a  greater  majority  of  the 
plants  fluctuating  around  a  mean  larger  in  magnitude  than  that  of  the  Fi 
plants.  This  deduction  is  confirmed  by  the  frequency  distributions  of  Table 
19.5,  as  both  of  these  populations  have  a  greater  percentage  of  their  plants 
in  lower  classes  of  the  frequency  distributions  than  does  the  Fi  population. 
Powers,  Locke,  and  Garrett  (1950)  have  shown  that  the  data  give  a  good  fit 
to  frequency  distributions  calculated  on  the  assumption  that  the  parents  are 
differentiated  by  three  pairs  of  genes. 

Here,  proof  of  recombination  of  genes  to  produce  plants  in  the  F2  and  Bi 
to  Ponderosa  populations  with  greater  weight  per  locule  than  Fi  plants  is 
fairly  conclusive.  Also,  since  the  number  of  gene  pairs  or  closely  linked  blocks 
of  genes  is  few,  it  should  be  possible  by  selection  to  establish  inbred  lines  re- 
taining this  advantage. 

MAIN  AND  COMPONENT  CHARACTERS 

The  data  from  the  parental  and  hybrid  populations  of  tomatoes  on  the 
main  and  component  characters  provide  information  concerning  the  rela- 
tions between  gene  recombination,  dominance,  and  heterosis. 


306 


LEROY  POWERS 


Weight  of  Fruit  and  Its  Component  Characters 

The  data  on  weight  per  locule,  number  of  locules,  and  weight  per  fruit  for 
the  Porter  X  Ponderosa  hybrid  and  parental  populations  are  given  in  Tables 
19.4  and  19.6.  On  the  arithmetic  scale,  smaller  numbers  of  locules  show  par- 
tial dominance.  On  the  logarithmic  scale  the  means  of  the  Fj  and  F2  popula- 
tions are  not  significantly  different  from  the  average  of  the  means  of  the 
Porter  and  Ponderosa  populations.  The  mean  of  the  Bi  to  Porter  population 
is  not  significantly  different  from  the  average  of  the  means  of  the  Porter 
and  Fi  populations.  The  mean  of  the  Bi  to  Ponderosa  population  is  not  sig- 

TABLE  19.6 

THE  ARITHMETIC  AND  LOGARITHMIC  MEANS  FOR  NUMBER  OF  LOCULES 

AND  WEIGHT  PER  FRUIT  OF  PORTER  X  PONDEROSA  TOMATO 

HYBRID  AND  PARENTAL  POPULATIONS* 


Number  of  Locules 

Weight  per  Fruit 

PoPUL.iTION 

Arith- 
metic 

Logarithmic 

Arith- 
metic 

Logarithmic 

Porter 

2.1 
3.1 
4.5 
4.7 
7.1 
10.0 

0.30707210.002151 
0.468411+0.008158 
0.637265  +  0.007663 
0.628793+0.012522 
0.82940410.007738 
0.983292  +  0.017094 

21.5 
36.6 
65.0 
63.5 
97.3 
97.7 

1.326101+0.012358 

Bi  to  Porter 

F: 

Fo 

Bi  to  Ponderosa .  . 
Ponderosa 

1.539833  +  0.010394 
1.806845  +  0.009416 
1.762614  +  0.013078 
1.95443010.013269 
1.965097  +  0.008750 

*  Grown  at  Woodward,  Oklahoma,  in  1941;  original  data  taken  in  numbers  and  grams  and  transformed  to 
logarithms  to  obtain  the  means  and  standard  errors  of  the  logarithms. 


nificantly  different  from  the  average  of  the  means  of  the  Fi  and  Ponderosa 
populations.  Hence,  on  the  logarithmic  scale  there  is  no  dominance,  and  the 
data  indicate  that  the  genetic  variability  follows  the  logarithmic  scale.  In 
other  words,  the  effects  of  the  genes  differentiating  weight  per  locule  are 
multiplicative.  This  is  true  of  both  the  intraallelic  and  interallelic  inter- 
actions. 

Thus  on  the  logarithmic  scale  number  of  locules  shows  no  dominance, 
weight  per  locule  shows  heterosis  (Table  19.4)  and  the  two  combine  addi- 
tively  to  give  weight  per  fruit.  For  weight  per  fruit  the  Fi  indicates  partial 
dominance  of  greater  weight  per  fruit,  the  Bi  to  Ponderosa  complete  domi- 
nance, and  the  Bi  to  Porter  no  dominance.  On  the  arithmetic  scale  the  two 
component  characters  unite  multiplicatively,  and  the  Fi  indicates  partial 
dominance  of  greater  weight  per  fruit,  the  Bi  to  Ponderosa  complete  domi- 
nance, and  the  Bi  to  Porter  partial  dominance  of  smaller  weight  per  fruit. 
Then  it  is  clear  that  regardless  of  scale,  one  of  the  component  characters 
shows  some  degree  of  dominance,  the  other  heterosis.  They  combine  to  pro- 
duce the  main  character  which  in  turn  shows  some  degree  of  dominance. 


GENE   RECOMBINATION   AND  HETEROSIS  307 

Powers,  Locke,  and  Clarrett  (1950)  found  the  number  of  major  gene  j)airs 
differentiating  number  of  locules  to  be  3.  Since  weight  per  locale  was  found 
to  be  differentiated  by  3  pairs  of  major  genes,  a  comparatively  few  (prob- 
ably 6)  pairs  of  major  genes  differentiate  weight  per  fruit.  Hence,  the  number 
of  major  gene  pairs  responsible  for  heterosis  of  weight  per  locule  is  no  greater 
than  the  number  of  major  gene  pairs  responsible  for  no  dominance  of  num- 
ber of  locules  and  partial  or  complete  dominance  of  weight  per  fruit  on  the 
logarithmic  scale.  Then,  in  this  study  the  number  of  pairs  of  major  genes 
differentiating  the  character  has  no  bearing  on  whether  the  hybrid  popula- 
tions will  show  no  dominance,  partial  dominance,  complete  dominance,  or 
heterosis. 

From  these  results  it  follows  that  in  this  material  recombination  of  genes 
to  retain  the  advantages  of  heterosis  is  no  different  than  recombination  of 
genes  to  combine  desirable  characters.  Furthermore,  these  data  furnish 
rather  convincing  evidence  that  dominance  and  heterosis  are  different  de- 
grees of  expression  of  the  same  physiological  genetic  phenomena,  as  was 
postulated  by  Powers  (1941,  1944). 

Main  and  Component  Characters  of  45  Hybrids  Produced 
by  Crossing   10  Inbred  Lines  of  Tomatoes 

Table  19.7  summarizes  the  dominance  relations  of  the  main  and  com- 
ponent characters  of  45  hybrids  produced  by  crossing  10  inbred  lines  of 
tomatoes. 

The  percentage  values  given  in  Table  19.7  were  calculated  from  data  pre- 
sented in  a  previous  article  (Powers,  1945).  The  reader  is  referred  to  this 
article  for  the  experimental  design,  a  description  of  the  material,  and  meth- 
ods. Here,  only  the  method  of  compiling  the  data  need  be  given.  All  of  the 
values  of  this  table  with  the  exception  of  those  listed  under  heterosis  were 
calculated  from  the  formula  100[2Fi/(Pi  +  A)].  The  percentages  listed 
under  the  column  headings  "heterosis"  were  calculated  from  the  formula 
lOO(Fi/Pi)  and  10G{Fi/F2),  respectively.  Fi  is  the  mean  of  the  Fi  popula- 
tion, Fi  the  mean  of  the  parent  with  the  smaller  value,  and  A  the  mean  of 
the  parent  with  the  larger  value.  The  11  characters  listed  in  Table  19.7  were 
originally  expressed  in  the  following  units  of  measurement:  Spread  of  plant 
in  inches,  yield  per  plant  in  grams,  number  of  fruit  that  ripened  per  plant, 
height  per  plant  in  inches,  weight  per  locule  of  the  fruit  in  grams,  number  of 
days  from  first  fruit  set  to  first  fruit  ripe,  number  of  days  from  first  bloom 
to  first  fruit  set,  weight  of  fruit  in  grams,  number  of  days  from  seeding  to 
first  bloom,  number  of  locules  per  fruit,  and  number  of  days  from  seeding 
to  first  fruit  ripe. 

The  odds  against  any  value  belonging  in  an  adjacent  classification  (col- 
umn) are  greater  than  19: 1  with  the  exception  of  the  two  values  designated 
with  an  asterisk.  Even  for  these  two  values  the  odds  against  their  deviating 


308 


LEROY  POWERS 


more  than  one  class  are  greater  than  19: 1.  When  interpreting  the  data  it  is 
necessary  to  have  in  mind  that  parental  percentage  values  would  have  fallen 
into  the  complete  dominance  columns,  the  Pi  value  into  the  first  such  col- 
umn, and  the  Pi  into  the  second  such  column.  Also,  it  should  be  kept  in  mind 
that  the  values  listed  in  Table  19.7  are  for  the  different  Fi  hybrids,  and  with 
the  exception  of  the  values  listed  under  the  columns  headed  "heterosis"  are 
percentages  based  on  the  means  of  the  two  respective  parents.  The  percent- 
ages listed  under  the  heterosis  columns  are  based  on  the  mean  of  the  parent 
that  fell  into  the  adjacent  complete  dominance  columns. 

TABLE  19.7 

PERCENTAGE  RANGE  IN  EXPRESSION  OF  DOMINANCE  FOR 
DIFFERENT  CHARACTERS  OF  Fi  TOMATO  HYBRIDS* 


Dominance 

Character 

Heter- 
osis 

Com- 
plete 

Partial 

None 

Partial 

Com- 
plete 

Heter- 
osis 

Spread  of  plant   .          .... 

114 
166 
172 
112 
109 

122 

Yield,  ripe  fruit  per  plant.  .  . 

106 
99 

100 
98 

99 

117 
142 
104 

103 

171 

Number,  ripe  fruit  per  plant. 

78 

96 
70 

93 

155 

Height  of  plant 

121 

Weight  per  locule 

119 

Period,  first  fruit  set  to  first 
fruit  ripe 

90 

80 

'95' 
69 

93 

Period,   first   bloom    to   first 
fruit  set      

75 
95 
89 
73 

95 

125* 

VVpie^ht  ner  fruit 

53 
99 
79 

96 

102 

100* 

96 

Period,  seeding  to  first  bloom . 

Number  of  locules  per  fruit.  . 

Period,  seeding  to  first  fruit 

ripe 

*  As  measured  by  percentages  of  averages  of  values  of  parents  and  percentages  of  parental  values. 

If  dominance  and  heterosis  are  different  degrees  of  expression  of  the  same 
physiological  genetic  phenomena,  then  the  different  genotypes,  as  represented 
by  the  different  Fi  hybrids,  might  be  expected  to  show  ranges  in  expression 
of  a  given  character  from  different  degrees  of  dominance  to  heterosis. 

Every  character  listed  in  Table  19.7  except  number  of  days  from  first  fruit 
set  to  first  fruit  ripe,  in  the  different  hybrids,  ranges  from  some  degree  of 
dominance  to  heterosis.  Yield  in  grams  of  ripe  fruit  per  plant,  depending 
upon  the  genotype  (Fi  hybrid),  varied  from  no  dominance  through  all  classes 
to  heterosis  for  increased  yield.  Number  of  ripe  fruit  per  plant  and  height  of 
plant  varied  through  all  classes  from  partial  dominance  of  a  decrease  in  mag- 
nitude of  these  two  characters  to  heterosis  for  an  increase.  Weight  of  fruit 
in  grams,  number  of  days  from  seeding  to  first  bloom,  and  number  of  locules 
per  fruit  varied  from  no  dominance  to  heterosis  for  a  decrease  of  these  char- 
acters. Considering  all  of  the  characters  there  is  a  continuous  array  of  values 


GENE  RECOMBINATION  AND  HETEROSIS  309 

(that  is  values  in  all  classes)  from  heterosis  for  decrease  of  a  character  to 
heterosis  for  increase  of  a  character,  depending  upon  the  character  and  geno- 
type (Fi  hybrid). 

The  most  logical  conclusion  from  these  figures  is  that  dominance  and  het- 
osis  to  a  considerable  extent  are  different  degrees  of  expression  of  the  same 
physiological  genetic  phenomena.  This  hypothesis  is  greatly  strengthened  by 
findings  of  Powers  (1941)  that  whether  a  character  shows  dominance  or  het- 
erosis in  some  cases  is  dependent  upon  the  environment  and  in  other  cases 
upon  the  genotype.  As  pointed  out  previously,  gene  recombination  in  rela- 
tion to  heterosis  is  no  different  from  combining  any  two  desirable  characters 
by  recombination  of  genes.  A  study  of  the  component  characters  of  the  main 
characters  given  in  Table  19.7  offers  further  evidence  in  support  of  this 
contention. 

Yield  of  ripe  fruit  as  determined  by  weight  of  fruit  in  grams  is  dependent 
upon  number  of  fruits  that  ripen  and  weight  per  fruit.  The  first  of  these 
component  characters,  depending  upon  the  Fi  hybrid  being  considered,  varies 
from  partial  dominance  of  fewer  number  of  ripe  fruits  to  heterosis  for  an  in- 
creased number  of  ripe  fruits.  The  second  component  character  varies  from 
no  dominance  to  heterosis  for  smaller  weight  per  fruit.  They  combine  mul- 
tiplicatively,  and  in  many  cases  result  in  heterosis  for  yield  of  fruit  (Table 
19.7  and  Powers,  1944).  Here  again,  then,  is  a  case  involving  combination 
of  characters  to  produce  heterosis.  To  retain  some  of  the  benefits  of  heterosis 
in  inbred  lines  would  involve  recombination  of  the  genes  differentiating  the 
two  component  characters. 

In  turn  the  number  of  fruit  that  ripens  is  dependent  to  a  large  extent  at 
Cheyenne,  Wyoming,  on  earliness  of  maturity,  number  of  days  from  seeding 
to  first  fruit  ripe  (Powers,  1945).  Earliness  of  maturity  varies  from  partial 
dominance  of  fewer  days  from  seeding  to  first  fruit  ripe  to  heterosis  for  the 
shorter  period.  The  component  characters  of  earliness  of  maturity  are  period 
from  seeding  to  first  bloom,  period  from  first  bloom  to  first  fruit  set,  and  pe- 
riod from  first  fruit  set  to  first  fruit  ripe.  Number  of  days  from  seeding  to 
first  bloom  varies  from  no  dominance  to  heterosis  for  the  shorter  period. 
Number  of  days  from  first  bloom  to  first  fruit  set  varies  from  complete  domi- 
nance of  the  longer  period  to  heterosis  of  the  shorter  period.  Number  of  days 
from  first  fruit  set  to  first  fruit  ripe  varies  from  partial  dominance  of  the  longer 
period  to  complete  dominance  of  the  shorter  period. 

Weight  per  fruit  is  dependent  upon  weight  per  locule  and  number  of 
locules  per  fruit.  Weight  per  locule  varies  from  partial  dominance  for  less 
weight  per  locule  to  heterosis  for  greater  weight  per  locule.  Number  of 
locules  varies  from  no  dominance  to  heterosis  for  fewer  locules.  On  the 
arithmetic  scale  these  two  component  characters  combine  multiplicatively 
so  that  weight  per  fruit  varies  from  no  dominance  to  heterosis  for  less  weight 
per  fruit. 


310  LEROY  POWERS 

From  the  above,  as  was  found  true  for  yield  per  plant,  the  heterosis  noted 
for  earliness  of  maturity  results  from  the  combination  of  component  charac- 
ters which  in  certain  Fi  hybrids  may  themselves  exhibit  heterosis.  The  same 
is  true  for  weight  per  fruit.  In  other  words,  the  study  of  genetics  of  heterosis 
has  been  somewhat  simplified  by  breaking  the  main  characters  down  into 
their  component  characters.  Also,  as  before,  the  study  shows  that  gene  re- 
combination to  retain  some  or  all  of  the  increase  of  the  Fi  hybrid  over  the 
parents  is  dependent  upon  the  same  physiological  genetic  phenomena  as  are 
involved  in  attempting  to  combine  two  or  more  desirable  characters  into  a 
single  inbred  line. 

RECOVERING  INBRED  LINES  RETAINING  ADVANTAGES 
ATTRIBUTABLE  TO  HETEROSIS 

The  physiological  genetic  phenomena  that  hinder  or  aid,  by  the  recombi- 
nation of  genes,  the  recovery  of  inbred  lines  retaining  some  or  all  of  the 
advantages  attributable  to  heterosis  are  the  same  as  those  emphasized  by 
Jones  (1917)  and  East  (1936).  These  are  the  number  of  gene  pairs  differen- 
tiating the  parents,  linkage  relations  of  the  genes,  pleiotropy,  and  the  inter- 
action of  the  genes  as  determined  by  the  measurement  of  end  products,  both 
interallelic  and  intraallelic.  This  genetic  information  can  be  obtained  only 
by  rather  detailed  genetic  studies.  With  the  quantitative  characters  such 
studies  are  expensive  and  time  consuming.  Hence,  very  few  such  studies  have 
been  made  with  tomato  hybrids.  Powers,  Locke,  and  Garrett  (1950)  and 
Powers  (1950b)  have  made  a  gene  analysis  for  some  of  the  main  characters  and 
their  more  obvious  components.  Even  though  the  gene  analysis  for  number 
of  days  from  seeding  to  first  fruit  ripe  has  been  completed  for  only  one  of 
the  four  crosses  to  be  considered,  this  character  and  weight  per  locule  will  be 
treated  as  component  characters  of  yield  of  ripe  fruit  per  plant  in  the  section 
dealing  with  number  of  pairs  of  genes  differentiating  the  parents. 

Number  of  Gene  Pairs  Differentiating  Parents 

In  considering  the  bearing  that  number  of  gene  pairs  differentiating  the 
parents  has  upon  gene  recombination  and  heterosis,  just  two  characters  will 
be  considered:  weight  per  locule  and  number  of  days  from  seeding  to  first 
fruit  ripe.  That  both  of  these  characters  have  an  effect  upon  yield  of  ripe  fruit 
should  be  kept  in  mind  during  the  analyses  and  discussions  which  follow. 
Also,  other  component  characters  listed  in  Table  19.7  could  be  studied.  How- 
ever, the  additional  information  gained  would  not  justify  the  time  and  space 
required,  as  the  fundamental  principles  involved  can  be  brought  out  from 
an  analysis  and  discussion  of  the  data  for  the  two  characters  chosen.  The 
number  of  gene  pairs  (effective  factors;  Mather,  1949)  differentiating 
weight  per  locule  has  been  determined  for  all  the  hybrid  populations  listed 
in  Table  19.8.  For  days  from  seeding  to  first  fruit  ripe  the  number  of  gene 


GENE  RECOMBINATION  AND  HETEROSIS 


311 


pairs  (effective  factors)  differentiating  the  parents  has  been  determined  for 
the  Porter  X  Ponderosa  hybrid  populations  only. 

In  discussing  the  bearing  the  number  of  gene  pairs  differentiating  the  two 
parents  has  upon  gene  recombination  and  heterosis,  information  concerning 
phenotypic  dominance  of  the  characters  for  the  hybrid  populations  is  neces- 
sary and  will  be  derived  by  studying  the  means  of  the  parental  and  hybrid 

TABLE  19.8 

MEANS  FOR  WEIGHT  PER  LOCULE  AND  NUMBER  OF  DAYS  FROM  SEED- 
ING TO  FIRST  FRUIT  RIPE  WITH  TYPE  AND  NUMBER  OF  GENE  PAIRS 
DIFFERENTIATING  THE  PARENTS  FOR  WEIGHT  PER  LOCULE* 


Danmark  X 

Danmark  X 

JOHANNISFEUER  X 

Porter  X 

Red  Currant 

JOHANNISFEUER 

Red  Currant 

Ponderosa 

Population 

No.  of 

No.  of 

No.  of 

No.  of 

Weight 

Days 

Weight 

Days 

Weight 

Days 

Weight 

Days 

per 

From 

per 

From 

per 

From 

per 

From 

Locule 

Seeding 

Locule 

Seeding 

Locule 

Seeding 

Locule 

Seeding 

(Gm.J 

to  Fruit 
Ripe 

(Gm.) 

to  Fruit 
Ripe 

(Gm.) 

to  Fruit 
Ripe 

(Gm.) 

to  Fruit 
Ripe 

Pit 

0.45 

156.9 

4.61 

164.9 

0.44 

126.0 

10.2 

147.7 

Bi  to  Pi 

0.97 

155.0 

6.72 

165,0 

1.04 

123.1 

11.8 

152.0 

F, 

2.33 

153.8 

7.96 

165,6 

2,70 

118.9 

14.4 

149.6 

F2 

2.12 

156.6 

8.35 

166.4 

2.12 

125.5 

13.5 

155.0 

Bi  toPz 

4.82 

159.7 

8.32 

167.6 

4.48 

124.7 

13.7 

168.8 

P2t 

10.36 

169.8 

9.92 

170.0 

6.20 

136,1 

9.8 

204.8 

Type  and  num- 

ber of  pairs 

of  genes 

Minor 

Major 



Minor 

Major 

Major 

40 -f 

2  or 

3 

40+ 
Major 
2  or 
3 

3 

8 

*  For  the  hybrid  populations  of  Danmark  X  Red  Currant,  Danmark  X  Johannisfeuer,  Johannisfeuer  X 
Red  Currant,  and  Porter  X  Ponderosa. 

t  Pi  is  Red  Currant,  Johannisfeuer,  Red  Currant,  and  Porter,  respectively. 
t  P2  is  Danmark,  Danmark,  Johannisfeuer,  and  Ponderosa,  respectively. 


populations  given  in  Table  19.8.  The  means  for  weight  per  locule  of  tomato 
fruits  and  number  of  days  from  seeding  to  first  fruit  ripe  together  with  the 
type  and  number  of  gene  pairs  differentiating  the  parents  for  weight  per 
locule  for  the  hybrid  populations  of  Danmark  X  Red  Currant,  Danmark  X 
Johannisfeuer,  Johannisfeuer  X  Red  Currant,  and  Porter  X  Ponderosa  are 
given  in  Table  19.8. 

The  first  two  hybrid  populations  were  grown  at  Cheyenne,  Wyoming,  in 
1938,  the  third  hybrid  population  at  the  same  location  in  1939,  and  the  last 
hybrid  whose  means  are  listed  in  the  extreme  right  hand  column  of  Table 
19.8  was  grown  at  Woodward,  Oklahoma,  in  1941.  The  means  of  this  table 


312  LEROY  POWERS 

were  taken  from  the  following  publications:  Powers  and  Lyon  (1941), 
Powers,  Locke,  and  Garrett  (1950),  and  Powers  (1950a).  The  data  will  be 
analyzed  to  obtain  information  concerning  the  recombination  of  the  genes 
differentiating  weight  per  locule  and  number  of  days  from  seeding  to  first 
fruit  ripe.  Also,  the  data  will  be  studied  to  ascertain  the  probable  bearing 
this  information  has  upon  the  production  of  inbred  lines,  by  gene  recombina- 
tion, that  retain  some  or  all  of  the  advantages  attributable  to  heterosis  of 
yield  of  ripe  fruit  per  plant  which  the  hybrid  populations  would  be  expected 
to  exhibit. 

On  the  arithmetic  scale  the  Danmark  X  Red  Currant  populations  show 
partial  phenotypic  dominance  for  smaller  weight  per  locule.  The  parents  of 
the  Danmark  X  Red  Currant  hybrid  were  found  to  be  differentiated  by  a 
large  number  of  gene  pairs  (probably  more  than  40)  which  individually  had 
minor  effects.  From  these  results  it  is  evident  that,  if  somewhere  near  one- 
half  of  the  genes  for  smaller  weight  per  locule  in  the  Danmark  X  Red  Cur- 
rant hybrid  populations  had  entered  the  cross  from  one  parent  and  the 
balance  from  the  other  parent,  smaller  weight  per  locule  would  have  shown 
heterosis.  Some  of  the  genes  must  be  linked  because  the  parents  have  a 
haploid  chromosome  number  of  12.  In  fact,  since  40  or  more  pairs  of  genes 
are  differentiating  the  parents,  it  seems  highly  probable  that  a  system  of 
linked  polygenes  is  involved.  With  40  pairs  of  genes  differentiating  the  par- 
ents in  the  F2,  to  recover  an  individual  possessing  all  of  the  genes  for  in- 
creased weight  per  locule  (without  linkage)  would  require  a  population  of 
lO^'*  individuals.  The  size  of  such  a  population  can  be  appreciated  by  con- 
sidering the  fact  that  10"  is  100  billion.  The  bearing  this  has  upon  the 
feasibility  of  recovering  from  segregating  populations  inbred  lines  retaining 
much  of  the  advantage  that  might  be  exhibited  by  Fi  hybrids  is  apparent. 

The  Red  Currant  parent  which  possesses  small  weight  per  locule  also 
possesses  earliness  of  maturity.  Hence,  some  of  the  genes  tending  to  increase 
weight  per  locule  are  almost  certain  to  be  located  on  the  same  chromosomes 
with  a  non-beneficial  gene  or  genes  tending  to  increase  the  time  required  for 
maturity.  However,  due  solely  to  the  large  number  of  gene  pairs  differentiat- 
ing weight  per  locule,  with  no  close  linkage,  pleiotropy,  or  unfavorable  in- 
terallelic  and  intraallelic  interactions  of  the  genes,  only  a  comparatively 
small  amount  of  the  increased  weight  per  locule  of  the  Danmark  parent  could 
be  combined  with  the  earliness  of  maturity  of  the  Red  Currant  parent  by 
selection  in  the  F2  or  backcross  populations. 

Weight  per  locule  and  earliness  of  maturity  have  a  material  influence  on 
yield  of  ripe  fruit  per  plant  (Powers,  1945).  In  some  crosses  (see  Tables  19.7 
and  19.8)  greater  weight  per  locule  is  at  least  partially  dominant.  Since  the 
shorter  period  for  days  from  seeding  to  first  fruit  ripe  for  the  Danmark  X 
Red  Currant  cross  shows  heterosis  (Table  19.8)  the  hybrid  populations  would 
be  expected  to  show  heterosis  for  yield  of  ripe  fruit  per  plant  in  crosses  hav- 


GENE  RECOMBINATION  AND  HETEROSIS  313 

ing  such  a  polygenic  system  conditioning  weight  per  locule,  provided  greater 
weight  per  locule  was  at  least  partially  dominant,  and  provided  the  genes 
for  increased  weight  per  locule  and  shorter  period  from  seeding  to  first  fruit 
ripe  were  divided  between  the  two  parents.  The  analyses  and  discussions  in 
the  immediately  preceding  paragraphs  show  that  in  such  an  event  it  would 
be  almost  impossible  to  obtain  inbred  lines  which  through  gene  recombina- 
tion would  retain  any  appreciable  amount  of  the  yield  of  the  Fi  hybrid. 

On  the  arithmetic  scale  the  Johannisfeuer  X  Red  Currant  populations 
show  partial  phenotypic  dominance  of  smaller  weight  per  locule  with  the 
exception  of  the  Bi  to  P2  which  indicates  no  dominance.  The  parents  of  the 
Johannisfeuer  X  Red  Currant  hybrid  populations  were  found  to  be  differ- 
entiated by  a  large  number  of  gene  pairs  (probably  more  than  40)  each  of 
which  individually  had  minor  effects  and  in  addition  by  a  few  gene  pairs 
(probably  2  or  3)  having  major  effects.  In  these  hybrid  populations  the  total 
effect  of  the  minor  genes  was  greater  than  the  total  effect  of  the  major  genes. 
Again  the  shorter  period  from  seeding  to  first  fruit  ripe  showed  heterosis. 

With  the  number  and  type  of  gene  pairs  conditioning  weight  per  locule 
found  for  the  Johannisfeuer  X  Red  Currant  hybrid,  and  provided  the  genes 
differentiating  weight  per  locule  exhibited  at  least  partial  dominance,  as  is 
indicated  for  the  Danmark  X  Johannisfeuer  populations,  certain  parental 
combinations  of  the  genes  would  result  in  the  hybrid  populations  showing 
heterosis  for  increased  yield  of  fruit  per  plant.  Since  comparatively  few  ma- 
jor gene  pairs  differentiate  weight  per  locule,  it  should  be  possible  by  re- 
combination of  genes  through  selection  in  F2  and  backcross  populations  of 
such  a  cross  to  combine  into  inbred  lines  some  of  the  increased  yield  at- 
tributable to  heterosis. 

The  Danmark  X  Johannisfeuer  hybrid  populations  show  partial  pheno- 
typic dominance  for  greater  weight  per  locule,  and  complete  dominance  for 
shorter  period  from  seeding  to  first  fruit  ripe.  Two  or  three  major  gene  pairs 
were  found  to  be  differentiating  weight  per  locule.  For  weight  per  locule  and 
number  of  days  from  seeding  to  first  fruit  ripe,  dominance  is  such  that  had  the 
genes  tending  to  increase  each  of  these  two  characters  been  divided  between 
the  two  parents,  the  hybrid  populations  would  have  shown  heterosis  for  both 
component  characters.  Likewise,  if  the  above  conditions  had  been  fulfilled, 
yield  of  ripe  fruit  per  plant  would  have  shown  heterosis  in  the  hybrid  popu- 
lations. 

The  Porter  X  Ponderosa  hybrid  populations  showed  at  least  partial  genie 
dominance  for  weight  per  locule  (Powers,  Locke,  and  Garrett,  1950).  The 
parents  were  found  to  be  differentiated  by  three  pairs  of  genes  and  the  genes 
tending  to  increase  weight  per  locule  were  distributed  between  the  two  par- 
ents. As  was  to  be  expected,  the  hybrid  populations  showed  heterosis  for  in- 
creased weight  per  locule.  Period  from  seeding  to  first  fruit  ripe  showed  al- 
most if  not  complete  dominance  for  the  shorter  period  from  seeding  to  first 


314  LEROY  POWERS 

fruit  ripe.  The  number  of  major  gene  pairs  found  to  be  differentiating  the 
parents  was  eight.  Due  to  the  magnitude  of  the  work  involved  it  was  not 
possible  to  measure  yield  of  fruit,  but  in  all  probability  the  hybrid  popula- 
tions of  this  cross  would  have  shown  heterosis  for  yield  of  ripe  fruit  per  plant. 
In  such  an  event  it  seems  highly  probable  that  some  and  perhaps  a  con- 
siderable amount  of  the  increase  in  yield  attributable  to  heterosis  could  be 
obtained  in  inbred  lines  through  recombination  of  genes. 

Considering  the  data  for  all  the  crosses  listed  in  Table  19.8  the  informa- 
tion may  be  summarized  as  follows:  In  the  Danmark  X  Red  Currant  cross 
a  large  number  of  gene  pairs  differentiates  the  parents  and  individually  the 
genes  have  minor  effects.  The  same  is  true  of  the  Johannisfeuer  X  Red  Cur- 
rant cross  with  the  exception  that  two  or  three  pairs  of  genes  have  major 
effects.  In  both  the  Danmark  X  Johannisfeuer  and  the  Porter  X  Ponderosa 
crosses  weight  per  locule  is  differentiated  by  a  comparatively  few  pairs  of 
genes  having  major  effects.  It  is  apparent  that  in  the  Porter  X  Ponderosa 
cross  it  should  be  possible  by  selection  in  the  segregating  populations  to  ob- 
tain by  recombination  of  genes  inbred  lines  equaling  if  not  excelling  the  Fi 
fruits  in  weight  per  locule. 

The  discussions  treating  weight  per  locule  and  number  of  days  from  seed- 
ing to  first  fruit  ripe  as  component  characters  of  yield  of  ripe  fruit  per  plant 
reveal  that  the  recombination  of  genes  to  retain  some  or  all  of  the  advantages 
of  the  Fi  hybrid  is  analogous  to  recombination  of  genes  for  the  purpose  of 
combining  desirable  characters. 

Linkage   Relations 

Linkage  may  be  an  aid  or  a  hindrance  to  gene  recombination.  The  data 
in  Table  19.9  were  computed  to  facilitate  a  consideration  of  the  manner  in 
which  different  linkage  relations  may  affect  recombination  of  genes. 

Certain  assumptions  were  essential  to  a  calculation  of  the  data.  First,  it 
was  assumed  that  the  coefficient  of  coincidence  is  1.  Since  in  most  cases  there 
is  interference,  to  assume  a  coefficient  of  coincidence  of  1  is  to  err  on  the 
conservative  side.  For  example,  all  the  values  given  in  the  second  row  head- 
ing (with  the  exception  of  the  first  and  last)  would  increase  as  the  coefficient 
of  coincidence  became  smaller.  The  reverse  is  true  of  the  figures  in  the  third 
and  fourth  columns.  The  frequencies  listed  in  the  second,  third,  and  fourth 
columns  of  Table  19.9  are  the  theoretical  number  of  individuals  in  the  F2  pop- 
ulation carrying  the  12  plus  genes  in  the  homozygous  condition.  The  cross- 
over values  expressed  as  decimal  fractions  are  assumed  to  be  equal  for  the 
different  sections  of  the  chromosomes  delimited  by  any  two  adjacent  genes. 

The  conclusions  to  be  drawn  from  the  theoretical  data  of  Table  19.9  are  not 
invalidated  by  these  assumptions.  They  merely  serve  the  purpose  of  allowing 
the  calculation  of  theoretical  values  for  illustrative  purposes.  Other  assump- 
tions such  as  different  values  of  crossing  over  for  the  various  sections  of  the 


GENE  RECOMBINATION  AND  HETEROSIS 


315 


chromosomes  and  diflferent  numbers  of  genes,  combinations  of  genes  in  the 
parents,  and  number  of  linkage  groups  would  not  alter  the  conclusions  to  be 
drawn.  In  the  illustration  chosen  only  two  linkage  groups  are  shown  and  each 
has  three  pairs  of  genes.  Also,  the  top  row  of  genes  represents  the  gamete  from 
one  parent  and  the  lower  row  of  genes  the  gamete  from  the  other  parent.  In 
all  three  assumed  cases,  3  plus  and  3  minus  genes  entered  the  cross  from 
each  parent. 

It  is  evident  that  innumerable  plausible  cases  could  be  assumed,  but  the 
fundamental  principles  derived  from  a  consideration  of  the  theoretical  values 
given  in  the  table  would  not  be  altered.  One  further  assumption  should  be 

TABLE  19.9 

THEORETICAL  NUMBER  OF  INDIVIDUALS  IN  THE  F2  POPU- 
LATION THAT  CARRY  12  PLUS  (+)  GENES  WHEN  THE  PAR- 
ENTS ARE  DIFFERENTIATED  BY  6  PAIRS  OF  GENES,  EACH 
OF  2  CHROMOSOME  PAIRS  CARRYING  3  PAIRS  OF  GENES* 


Cross- 

Linxage Relations  in  Fi  (Number  per  Million) 

over 
Value 

(^^^)(;;;) 

(^i;)(;;^) 

(^;i)(;i;) 

0.000... 
0.075... 
0.225... 
0.375... 
0.450... 
0.500... 

62,500 

33,498 

8,134 

1,455 

523 

244 

0.000 

1.448 

57.787 

188.596 

234.520 

244.141 

0.000000 

0.000063 

0.410526 

24.441630 

105.094534 

244.140625 

*  The  crossover  values  for  each  section  of  the  chromosome  being  equal  and  of  the  magni- 
tude shown. 

mentioned.  In  every  case  the  plus  genes  are  assumed  to  give  an  increase  in 
some  desirable  quantitative  character  and,  comparatively,  the  minus  genes 
a  decrease.  Finally,  in  the  table  two  extreme  situations  are  shown,  namely 
that  in  which  there  is  no  crossing  over  and  that  in  which  the  two  sections 
of  the  chromosome  between  adjacent  genes  show  50  per  cent  of  crossing  over. 

The  data  in  the  second  column  apply  to  that  situation  in  which  all  of  the 
plus  genes  occur  in  one  member  of  the  homologous  chromosomes  in  each  of 
the  two  pairs  of  chromosomes  depicted.  In  the  case  of  50  per  cent  of  crossing 
over  or  independent  inheritance,  only  244  individuals  in  a  million  of  the  F2 
population  possess  all  twelve  plus  genes.  The  number  of  such  individuals 
among  a  million  F2  individuals  increases  with  a  decrease  in  the  percentage 
of  crossing  over  until  with  no  crossing  over  62,500  individuals  in  a  million 
possess  all  six  pairs  of  the  plus  genes  in  the  homozygous  condition. 

The  data  in  the  third  column  apply  to  that  situation  in  which  two  plus 
genes  are  linked  with  one  minus  gene  in  one  member  of  a  chromosome  pair 
and  two  minus  genes  with  one  plus  gene  in  the  other  member  of  the  same 


316  LEROY  POWERS 

chromosome  pair.  In  this  column  the  situation  is  reversed  as  compared  to 
column  two.  Again  50  per  cent  of  crossing  over  gives  244  individuals  among 
a  million  in  the  F2  possessing  all  twelve  plus  genes.  This  decreases  with  a  de- 
crease in  the  percentage  of  crossing  over  until  with  no  crossing  over  no  indi- 
viduals in  the  infinite  F2  population  contain  more  than  eight  plus  genes. 
However,  since  two  of  the  plus  genes  are  carried  on  the  same  chromosome 
in  each  of  the  two  linkage  groups,  an  increase  in  the  linkage  intensity  results 
in  an  increased  number  of  individuals  in  the  F2  population  possessing  all 
eight  plus  genes  in  the  homozygous  condition. 

Here,  then,  is  a  case  in  which  close  linkage  facilitates  recombination  of 
desired  genes  up  to  a  certain  number,  and  from  a  practical  standpoint  further 
advances  by  selection  in  that  generation  are  impossible.  Also,  it  would  be 
difhcult  to  make  further  advances  by  continued  selection  in  later  genera- 
tions. In  the  F2  population  with  a  crossover  value  of  0.075  the  frequency  of 

the    ( T  T  _  )  ( T  T  _  )  genotype   e.xpressed   as   a   decimal   fraction   is 

0.183024  and  of  the  (t  t  7)(i  t  l)  genotype  is  0.014840. 

To  obtain  some  F3  families  derived  from  F2  plants  of  the  latter  genotype 
would  require  growing  at  least  300  selections  in  the  F3  generation.  To  sepa- 
rate the  F3  families  derived  from  the  F2  plants  of  the  former  genotype  from 
those  derived  from  the  latter  genotype  would  require  an  adequately  replicated, 
well  designed  experiment.  Anyone  who  has  worked  with  the  quantitative  char- 
acters either  in  genetics  or  plant  breeding  realizes  the  difficulties  besetting  such 
a  task.  After  such  F3  families  had  been  determined,  only  25  per  cent  of  the  in- 
dividuals would  be  of  the  (TTT)(TT_)  genotype.  These  would  have 
to  be  tested  in  the  F4  to  separate  them  from  F4  families  derived  from  F3  plants 
of  the  (i  i  7)  (i  t  l)  and  the  (^  ]j]  ~)  (^  ][]  ~)  genotypes.  Even 

with  the  small  number  of  genes  assumed  in  the  above  example,  it  would  not  be 
a  simple  matter  to  make  progress  by  continued  selection  in  later  generations. 
The  addition  of  a  few  more  genes  having  the  plus  and  minus  genes  alternating 
on  the  same  chromosome  would  make  further  progress  by  continued  selection 
in  generations  later  than  the  F2  practically  impossible.  From  the  above  it  is 
apparent  that  any  series  of  plus  genes  being  adjacent  without  minus  genes  in- 
tervening would  facilitate  recombination  of  desirable  genes  in  the  F2  genera- 
tion. It  seems  that  in  actual  genetic  and  plant  breeding  materials  many  such 
combinations  do  exist. 

The  figures  in  the  fourth  column  of  Table  19.9  are  the  theoretical  fre- 
quency distributions  for  that  situation  in  which  the  plus  and  minus  genes 
alternate  on  the  chromosome.  Again  the  number  of  individuals  expected  in 
the  F2  generations  possessing  all  twelve  plus  genes  decreases  rather  rapidly 
with  a  decrease  in  the  percentage  of  crossing  over.  Even  in  the  case  of  50  per 


GENE   RECOMBINATION  AND  HETEROSIS  317 

cent  of  crossing  over  it  is  doubtful  whether  it  is  possible  for  the  plant  breeder 
or  geneticist  to  isolate  individuals  from  the  F2  population  carrying  twelve 
plus  genes. 

The  data  in  Table  19.9  emphasize  that  even  with  the  probably  over- 
simplified genetic  situation  depicted  it  is  not  possible  to  recover  in  a  single 
individual  all  of  the  genes  favorable  to  the  production  of  a  desirable  charac- 
ter for  which  the  F2  population  is  segregating,  unless  the  favorable  genes  are 
located  on  the  same  chromosome  and  immediately  adjacent  to  each  other 
without  unfavorable  genes  intervening.  If  any  of  the  favorable  genes  are 
adjacent  to  each  other  without  unfavorable  genes  intervening,  then  decided 
advances  can  be  made  by  selection  in  the  F2  populations  up  to  a  certain  point. 
Beyond  that  point  further  selection  in  the  F2  will  have  no  effect,  and  selec- 
tion in  advanced  generations  does  not  offer  much  promise.  The  most  difficult 
situation  is  that  in  which  the  linkage  relation  is  such  that  the  favorable  and 
unfavorable  genes  alternate  on  the  chromosome  and  the  number  of  such 
linkage  groups  is  at  a  minimum  for  the  number  of  gene  pairs  involved. 

For  the  sake  of  clarity  of  illustration  only  three  linkage  relations  were 
shown.  However,  it  is  apparent  that  undoubtedly  in  the  material  available 
to  plant  breeders  and  geneticists,  the  possible  different  kinds  of  linkage  rela- 
tions are  almost  innumerable.  Some  will  aid  the  investigator  in  obtaining  the 
desired  recombination  of  genes  and  others  will  be  a  decided  hindrance.  In 
the  cases  of  undesirable  linkage  relations  it  will  be  almost  impossible  for  the 
breeder  to  obtain  individuals  possessing  recombinations  of  genes  making 
that  individual  equal  to  or  superior  to  the  Fi  for  the  character  exhibiting 
heterosis.  On  the  other  hand,  desirable  linkage  relations  may  make  it  pos- 
sible to  obtain  the  recombination  of  genes  sought  even  though  a  large  num- 
ber of  gene  pairs  differentiates  the  parents  used  in  hybridization. 

Pleiotropy,  and  Interallelic  and  Intraallelic  Interactions 

Powers,  Locke,  and  Garrett  (1950)  have  made  a  rather  detailed  genetic 
study  of  eight  quantitative  characters  in  hybrid  and  parental  populations 
involving  the  Porter  and  Ponderosa  varieties  of  Lycopersicon  esculentum 
Mill.  The  characters  studied  and  the  indicated  number  of  major  gene  pairs 
differentiating  the  parents  are  as  listed  immediately  below. 

Character  Gene  Symbols 

Percentage  of  flowers  that  set  fruit FifiFifiFifiFifi 

Period  from  seeding  to  first  fruit  ripe: 

Seeding  to  first  bloom BihiB-ibiBibz 

First  bloom  to  first  fruit  set SiS\S2Si,SiSi 

First  fruit  set  to  first  fruit  ripe RxfiRiTi 

Weight  per  fruit: 

Number  of  locules LcilciLcildLcilci 

Weight  per  locule WiWiWiW'zWzWz 

With  most  quantitative  characters  it  is  difficult  to  distinguish  between 
pleiotropy  and  linkage.  It  seems  highly  probable  that  linkage  instead  of  plei- 


318  LEROY  POWERS 

otropy  produced  the  relations  noted  by  the  above  authors  between  the  four 
series  of  genes  F/,  Ss,  Rr,  and  Lclc  with  the  exception  of  the  Ff  and  Ss  re- 
lation, because  all  the  associations  noted  are  those  expected  on  the  basis  of 
linkage.  If  pleiotropy  were  involved,  such  relations  would  be  coincidental, 
which  for  all  these  gene  series  is  highly  improbable.  However,  as  pointed  out 
by  Powers,  Locke,  and  Garrett  (1950)  some  of  the  genes  of  the  Ff  and  Ss 
series  must  be  identical,  as  percentage  of  flowers  that  set  fruit  has  an  effect 
on  period  from  first  bloom  to  first  fruit  set.  The  Lclc  and  Ww  series  of  genes, 
differentiating  number  of  locules  and  weight  per  locule,  respectively,  were 
independent  as  regards  linkage  and  pleiotropy.  In  these  studies  pleiotropy 
was  not  of  major  importance. 

Phenotypic  and  genie  dominance  furnish  some  information  concerning 
the  interallelic  and  intraallelic  interactions  of  the  genes.  That  genie  domi- 
nance is  dependent  upon  the  genotypic  milieu  was  pointed  out  by  Fisher 
(1931)  and  many  others  (Dobzhansky,  1941).  Hence  both  interallelic  and 
intraallelic  interactions  as  measured  by  end  products  are  second  order  inter- 
actions, genes  X  genes  X  the  environment. 

Any  of  the  interactions  of  genes  noted  as  affecting  any  of  the  component 
characters  dealt  with  in  the  study  by  Powers,  Locke,  and  Garrett  (1950) 
were  interactions  of  genes  differentiating  yield  of  ripe  fruit  per  plant.  With 
this  fact  in  mind,  it  is  interesting  to  note  the  interactions  of  the  genes  differ- 
entiating the  component  characters.  The  intraallelic  and  interallelic  interac- 
tions of  the  Ff  gene  series  were  such  that  genie  dominance  was  intermediate. 
The  intraallelic  and  interallelic  interactions  of  the  Bb  series  of  genes  were 
such  that  one  of  the  six  dominant  genes  shortened  the  period  from  seeding  to 
first  bloom  as  much  as  all  six,  which  shows  that  both  dominance  and  epistasis 
were  complete.  For  the  5^  series  and  Rr  series  of  genes,  genie  dominance  was 
complete.  Also,  the  effects  of  the  gene  pairs  were  cumulative. 

Had  the  dominant  genes  of  the  Ss  series  entered  the  cross  from  one  parent 
and  the  dominant  genes  from  the  Rr  series  entered  the  cross  from  the  other 
parent,  the  Fi  hybrid  would  have  shown  heterosis  for  earliness  of  maturity. 
Porter  would  then  represent  an  inbred  line  which  by  recombination  of 
genes  retained  the  earliness  of  maturity  of  the  Fi  hybrid.  Genie  dominance 
was  partial  for  genes  (LciLc^)  tending  to  produce  fewer  locules  per  fruit  and 
for  the  (Lcs)  tending  to  produce  more  locules  per  fruit.  A  series  of  genes  such 
as  Lci  and  Lc2,  some  entering  the  cross  from  one  parent  and  some  from  the 
other,  would  produce  an  Fi  hybrid  showing  heterosis  for  fewer  locules  per 
fruit.  On  the  other  hand  a  series  of  genes  such  as  (Lcs),  some  entering  the 
cross  from  one  parent  and  some  from  the  other,  would  produce  heterosis  for 
more  locules  per  fruit. 

Finally,  for  the  Wiv  series  of  genes,  genie  dominance  was  partial  for  in- 
creased weight  per  locule  and  the  effects  of  the  gene  pairs  were  cumulative. 
As  regards  this  character,  both  parents  did  contribute  genes  for  increased 


GENE  RECOMBINATION  AND  HETEROSIS  319 

weight  per  locule,  and  the  Fj  hybrid  did  show  heterosis  for  increased  weight. 
Also,  as  has  been  shown  in  the  F2  and  Bi  and  P2  populations  some  individuals 
were  obtained  having  greater  weight  per  locule  than  the  Fi  plants  and  this 
greater  weight  per  locule  proved  to  be  due  to  recombination  of  favorable 
genes. 

Also,  the  interallelic  interactions  of  the  genes  as  determined  by  the  inter- 
relations of  the  component  characters  are  of  interest  because  of  the  informa- 
tion they  provide  concerning  recombination  of  genes  and  heterosis.  The 
effects  of  the  Bb  series  of  genes,  the  ^'^  series,  and  the  Rr  series,  respectively, 
were  found  to  be  cumulative.  On  an  average  the  S  genes  would  be  e.xpected 
to  shorten  the  period  from  first  bloom  to  first  fruit  set  less  in  the  presence 
of  the  R  genes  than  in  the  presence  of  the  r  genes — if  the  physiological  reac- 
tions affecting  these  two  component  characters  that  were  instigated  by  the 
environment  were  the  same  as  those  instigated  by  the  6'5  and  Rr  gene  series. 
That  such  was  the  case  seems  probable  from  the  results  of  Goldschmidt's 
work  (1938)  with  phenocopies.  In  fact  it  seems  almost  axiomatic  that  this  was 
the  case,  because  the  second  order  interaction  (Ss  gene  series  X  Rr  gene 
series  X  environment)  was  such  that,  on  an  average,  when  the  ^5  series 
responded  to  a  given  environment  by  shortening  the  period  from  first  bloom 
to  first  fruit  set  the  Rr  series  in  the  same  plant  tended  to  produce  a  longer 
period  from  first  fruit  set  to  first  fruit  ripe.  Then  the  effects  of  these  two 
series  of  genes  were  less  than  additive  as  regards  the  dependent  character  pe- 
riod from  seeding  to  first  fruit  ripe 

About  the  same  situation  existed  in  respect  to  the  Lclc  series  and  the  Ww 
series  of  genes  in  that  greater  number  of  locules,  on  an  average,  was  ac- 
companied by  less  weight  per  locule.  This  type  of  interallelic  interaction 
would  tend  to  decrease  the  possibility  of  obtaining  inbred  lines  combining 
desirable  characters.  This  would  be  particularly  true  of  the  interallelic  inter- 
action between  the  Ss  and  Rr  gene  series,  because  a  shorter  period  from  first 
bloom  to  first  fruit  set  tended  to  be  accompanied  by  a  longer  period  from 
first  fruit  set  to  first  fruit  ripe 

The  data  do  not  furnish  any  evidence  concerning  that  type  of  intraallelic 
interaction  postulated  by  East's  (1936)  physiological  hypothesis,  other  than 
to  say  that  no  cases  of  overdominance  were  found.  This  would  indicate  that 
probably  overdominance  does  not  play  a  predominant  part  in  the  produc- 
tion of  heterosis  in  the  tomato  hybrids  studied. 


A.  J.  MANGELSDORF 

Experiment  Sfation,  Hawaiian  Sugar  Planters  Assodaiion ,  Honolulu,  T.H. 


Chapter  20 


Gene  Interaction  in  Heterosis 


Sugar  cane  behaves  very  much  like  corn  in  its  reaction  toward  inbreeding 
and  outcrossing.  Although  the  sugar  cane  flower  is  normally  provided  with 
both  male  and  female  organs,  male  sterility  is  not  uncommon.  Among  the 
varieties  that  produce  an  abundance  of  pollen,  many  are  partially  or  highly 
self-sterile.  As  a  consequence,  cross-fertilization  by  wind-borne  pollen  is  the 
rule  in  sugar  cane,  as  in  corn.  When  sugar  cane  is  subjected  to  self-pollina- 
tion, the  usual  result  is  a  reduction  in  seed  setting  and  a  marked  reduction  in 
the  vigor  of  the  offspring. 

The  sugar  cane  breeder  enjoys  one  great  advantage  over  the  corn  breeder: 
sugar  cane  can  be  propagated  asexually.  Each  node  on  the  stalk  is  provided 
with  a  bud  and  with  a  number  of  root  primordia.  In  field  practice,  stalks  of 
the  selected  variety  of  sugar  cane  are  sectioned  into  cuttings  of  two  or  more 
internodes  each.  These  cuttings  are  then  placed  horizontally  in  furrows  and 
covered  lightly  with  soil.  In  due  course  the  cutting  sends  out  its  roots,  the 
buds  develop  into  shoots,  and  a  new  plant  is  established. 

Were  it  possible  to  apply  this  procedure  to  corn,  and  thus  to  perpetuate 
outstanding  individuals  from  whatever  source,  it  is  unlikely  that  the  corn 
breeder  would  have  felt  obliged  to  resort  to  the  laborious  procedures  now 
employed. 

When  sugar  cane  varieties  are  propagated  by  cuttings,  the  traits  by  which 
we  are  able  to  distinguish  one  variety  from  another  maintain  their  integrity 
through  many  cycles  of  clonal  propagation.  This  is  true  not  only  of  morpho- 
logical traits,  but  also  of  physiological  traits. 

Sugar  cane  has  a  number  of  relatives  growing  in  the  wild,  some  of  which 
may  be  ancestral  to  the  original  cultivated  forms.  Wild  Saccharums  are  wide- 
ly distributed  in  the  tropical  and  sub-tropical  regions  of  the  Old  World,  from 
central  Africa  through  Asia  and  Malaya,  to  and  including  the  Indonesian 
and  many  of  the  more  westerly  Pacific  islands.  This  heterogeneous  array  of 

320 


GENE  INTERACTION   IN   HETEROSIS  321 

wild  forms  has  been  somewhat  arbitrarily  classified  into  two  great  groups — - 
the  S.  spontaneum  group  and  the  S.  robiistum  group.  Each  of  these  groups 
comprises  a  diversity  of  tyj)es  which  differ  among  themselves  in  morphology 
and  in  chromosome  number.  The  members  of  the  sponlanenni  group  have 
slender  stalks;  they  are  often  strongly  stoloniferous.  The  members  of  the 
robust U7n  group  have  hard,  woody  stalks,  sometimes  of  good  diameter;  sto- 
lons, if  present,  are  not  strongly  developed. 

The  original  cultivated  varieties  likewise  may  be  classified  into  two  great 
groups.  The  first  of  these  comprises  a  number  of  slender  varieties  which  ap- 
pear to  be  indigenous  to  India,  and  which  have  been  lumped  together  under 
the  name  S.  Barberi.  Certain  of  the  Barberi  varieties  bear  a  striking  resem- 
blance to  the  wild  spontaneums  of  that  region. 

The  New  Guinea  region  is  the  home  of  a  group  of  large-stalked  tropical 
cultivated  varieties  of  the  type  which  Linnaeus  named  S.  officinarum.  The 
wild  form  most  closely  resembling  S.  officinarum  and  possibly  ancestral  to 
it  is  S.  robustum,  which  is  indigenous  to  that  region. 

In  the  closely  related  genus  Sorghum,  the  difference  between  varieties 
having  pithy  stalks  containing  but  little  sugar,  and  varieties  with  sweet 
juicy  stalks,  has  been  shown  to  be  determined  by  a  single  major  gene.  In 
Saccharum  the  change  from  the  dry,  pithy,  low-sucrose  stalks  of  the  wild 
forms  to  the  juicy,  high-sucrose  stalks  of  the  cultivated  varieties  appears  to 
have  been  brought  about  by  several,  but  perhaps  by  no  more  than  three  or 
four  major  gene  changes. 

The  cultivated  and  wild  forms  also  differ  in  genes  for  stalk  size.  In  crosses 
between  the  two,  the  genes  responsible  for  the  slenderness  of  the  wild  forms 
show  a  high  degree  of  dominance. 

A  striking  feature  of  this  multiform  genus  is  the  prevalence  of  inter- 
fertility  among  its  members.  Widely  divergent  forms  can  be  crossed  without 
undue  difficulty.  The  resulting  hybrids  are  rarely  completely  sterile;  they 
are  often  highly  fertile.  The  explanation  is  presumably  to  be  sought  in  the 
polyploidy  which  is  characteristic  of  both  the  wild  and  the  cultivated  forms. 
They  range  in  chromosome  number  from  24  to  80  or  more  pairs.  It  appears 
that  once  the  minimum  chromosomal  complement  needed  to  produce  a  func- 
tional zygote  has  been  supplied,  there  is  considerable  latitude  in  the  number 
and  in  the  assortment  of  chromosomes  that  can  be  added  without  impairing 
the  viability,  or  even  the  fertility  of  the  hybrids. 

Since  the  breeder  is  as  yet  unable  to  create  superior  genes  at  will,  he  is 
obliged  to  content  himself  with  developing  new  combinations  of  the  genes 
available  in  whatever  breeding  material  he  may  be  able  to  assemble.  The 
sugar  cane  breeder  is  fortunate  in  having  in  the  wild  relatives  of  sugar  cane  a 
reservoir  of  genes  for  disease-resistance  and  hardiness.  Those  are  traits  that 
had  to  some  degree  been  lost  in  the  course  of  domestication.  Considerable 


322  A.  J.  MANGELSDORF 

use  has  already  been  made  of  the  wild  forms.  The  important  varieties  today 
are  almost  without  exception  complex  hybrids  that  include  in  their  ancestry 
representatives  of  both  the  S.  officinarum  and  the  S.  Barberi  groups  of  cul- 
tivated varieties,  together  with  representatives  of  one  or  both  of  the  wild 
species. 

Thus  the  sugar  cane  breeder  has  been  exploiting,  to  the  best  of  his  ability, 
the  advantages  that  heterosis  has  to  offer.  He  is,  however,  acutely  aware  that 
a  better  understanding  of  the  genetic  basis  of  heterosis  is  prerequisite  to  its 
more  effective  utilization.  Since  he  suffers  the  disadvantage  of  isolation  from 
the  centers  of  research,  he  cherishes  such  rare  opportunities  as  he  may  have 
to  peer  over  the  shoulder  of  the  research  worker,  to  whom  he  must  look  for 
new  facts  that  may  lead  to  a  better  understanding  of  the  mechanism  of  gene 
action  and  thus,  of  heterosis. 

Recently  some  of  us  who  are  engaged  in  sugar  cane  breeding  in  Hawaii 
formulated  a  number  of  postulates  with  the  object  of  providing  a  basis  for 
discussing  heterosis  and  related  matters.  These  postulates  have  been  ex- 
cerpted or  inferred  from  the  published  literature  and  from  correspondence 
with  workers  engaged  in  genetic  research,  whose  helpful  suggestions  are 
gratefully  acknowledged. 

Although  the  evidence  supporting  these  postulates  is  sometimes  meager, 
and  sometimes  capable  of  other  interpretations,  we  have  deliberately  phrased 
them  in  a  categorical  vein  in  the  belief  that  they  might  thus  better  serve 
their  primary  purpose — that  of  provoking  a  free  exchange  of  ideas. 

POSTULATES  RELATING  TO  INCIDENCE  OF  LESS  FAVORABLE  ALLELES 

1.  Naturally  self-fertilized  populations  tend  to  keep  their  chromosomes 
purged  of  all  alleles  other  than  those  which  in  the  homozygous  condition 
interact  to  best  advantage  with  the  remainder  of  the  genotype  and  with  the 
existing  environment^  to  promote  the  result  favored  by  natural  selection  (or 
by  human  selection).  This  does  not  imply  that  any  single  population  will  con- 
tain all  of  the  best  alleles  existing  in  the  species.  Selection  can  make  a  choice 
only  between  the  alleles  present  in  the  population. 

2.  In  addition  to  their  prevailing  (normal,  plus,  or  wild  type)  alleles,  cross- 
fertilized  organisms  such  as  corn  and  sugar  cane  carry  in  the  heterozygous 
condition,  at  many  loci,  recessive  alleles  which  in  the  homozygous  condition 
would  be  inferior  in  their  action  to  that  of  their  normal  or  prevailing  partners. 

3.  These  less  favorable  alleles  may  be  thought  of  as  belonging  to  one  of  two 
classes,  which,  although  differing  in  their  past  history,  may  have  similar 
physiological  consequences:  (a)  fortuitous,  resulting  from  sporadic  mutation, 
and  representing  the  errors  in  the  "trial  and  error"  of  the  evolutionary  proc- 
ess; or  {b)  relic,  representing  the  residue  of  what  were  once  the  prevailing 

1.  The  term  environment  is  here  used  in  a  broad  sense  to  mean  the  sum-total  of  the  ex- 
ternal influences  acting  upon  the  organism,  including  its  nutrition. 


GENE  INTERACTION  IN  HETEROSIS  323 

alleles  but  which,  in  the  course  of  evolution  or  under  a  changed  environment, 
have  been  displaced,  to  a  greater  or  lesser  degree,  by  still  better  alleles. 

4.  The  prevailing  allele  at  a  given  locus  has  reached  its  pre-eminent  posi- 
tion through  the  sifting  action  of  natural  selection  over  many  generations. 
Given  a  stable  environment,  further  improvement,  through  mutation,  at  that 
locus  would  long  since  have  materialized  if  the  chances  for  such  improvement 
were  high.  It  is  not  strange  that  random  mutation  should  only  rarely  be  able 
to  produce  a  superior  new  allele.  Nevertheless,  once  the  possibilities  for  im- 
provement through  recombination  of  existing  genes  have  been  exhausted, 
further  evolutionary  progress  will  be  contingent  upon  just  such  an  event, 
however  rare  its  occurrence  may  be. 

5.  Whether  dominant  or  recessive,  and  whether  in  a  naturally  self-ferti- 
lized or  naturally  cross-fertilized  population,  a  substantially  superior  mutant, 
once  established  in  the  population,  is  destined  to  increase  in  frequency  and  to 
become  the  prevailing  allele  in  the  population. 

6.  A  deleterious  dominant  is  doomed  to  eventual  extinction.  In  a  cross- 
breeding population  of  sufficient  size  a  deleterious  recessive  may  persist  in- 
definitely, its  incidence,  except  for  random  drift,  being  determined  by  the 
balance  between  its  elimination  by  selection  and  the  rate  at  which  it  recurs  by 
mutation. 

7.  The  best  allele  for  one  environment  may  not  be  best  for  another  envi- 
ronment. The  burden  of  less  favorable  alleles  which  cross-fertilized  organisms 
carry  along  generation  after  generation  is  not  an  unmitigated  liability.  It 
serves  as  a  form  of  insurance  by  providing  a  reservoir  of  adaptability  to 
changing  conditions. 

ROLE  OF  LESS  FAVORABLE  ALLELES 

Turning  now  to  the  role  of  these  less  favorable  alleles  in  the  heterosis 
phenomenon  as  manifested  in  naturally  cross-fertilized  organisms  we  may 
formulate  a  second  group  of  postulates: 

1.  At  many  and  perhaps  at  most  loci,  .la  is  as  good  or  nearly  as  good  as 
A  A,  and  both  A  A  and  A  a  are  better  than  aa. 

2.  There  may  be  a  few  loci  where  aa  is  better  than  .LI  or  .la.  This  is  par- 
ticularly likely  to  be  the  case  for  loci  affecting  traits  which  are  advantageous 
under  domestication,  but  disadvantageous  in  the  wild  under  natural  selec- 
tion. 

3.  There  may,  for  all  we  know,  be  occasional  loci  where  .LI'  is  better  than 
A  A  or  A' A'  (overdominance). 

4.  There  may  be  many  regions  in  the  chromosomes  which  behave  as  though 
A  A'  were  better  than  /L4  or  A' A'.  With  deleterious  recessive  alleles  in  the 
heterozygous  condition  at  many  loci,  it  seems  almost  inevitable  that  some  of 
these  will  be  closely  linked  in  the  repulsion  phase,  as  for  example  Ab/aB, 
which  in  the  absence  of  crossing  over  would  behave  as  a  single  locus,  the 


324  A.  J.  MANGELSDORF 

heterozygous  condition  of  which  is  superior  to  either  homozygote.  It  is  to  be 
expected  that  such  a  linkage  will  eventually  be  broken.  However,  there  may 
be  regions  in  the  chromosomes,  such  as  the  centromere  region,  for  example, 
where  crossing  over  is  reduced,  and  where  a  group  of  genes  may  act  indefi- 
nitely as  a  single  gene.  We  may  for  convenience  designate  the  effect  of  such 
reciprocal  apposition  of  favorable  dominants  to  their  less  favorable  reces- 
sives  as  a  pseudo-overdominance  effect.  It  will  be  noted  that  such  a  balanced 
defective  situation  conforms  with  the  dominance  and  linkage  hypothesis  ad- 
vanced by  Jones  as  an  explanation  of  the  heterosis  phenomenon. 

5.  Even  in  the  absence  of  linkage,  an  overdominance  type  of  reaction  (but 
resulting  from  pseudo-overdominance)  must  assert  itself  whenever  each  of 
the  two  members  of  a  pair  of  gametes  is  able  to  supply  the  favorable  domi- 
nant alleles  required  to  counteract  the  less  favorable  recessives  carried  by 
the  other  member  of  the  pair.  The  likelihood  of  success  in  retaining,  in  suc- 
cessive generations  of  selfing,  all  of  the  favorable  dominants  heterozygous 
in  Fi,  and  eliminating  all  of  the  less  favorable  recessives,  diminishes  ex- 
ponentially with  increasing  numbers  of  loci  heterozygous  in  Fi.  It  would 
seem  that  naturally  cross-fertilized  organisms  which  carry,  at  many  loci, 
deleterious  recessives  of  low  per  locus  frequency  in  the  population  could 
hardly  fail  to  manifest  a  pseudo-overdominance  type  of  response  to  inbreed- 
ing and  outcrossing. 

6.  From  an  evolutionary  standpoint,  it  may  be  important  to  distinguish 
between  the  consequences  of  (a)  true  overdominance  (heterozygosis  at  the 
locus  level)  and  {b)  pseudo-overdominance  (heterozygosis  at  the  zygote  level 
resulting  from  the  reciprocal  masking  of  deleterious  recessives  by  their 
dominant  alleles).  From  the  standpoint  of  the  breeder  who  is  of  necessity 
working  against  time,  this  distinction  may  have  little  practical  importance 
if  many  loci  are  involved  in  the  pseudo-overdominance  effect.  A  breeding 
plan  designed  to  deal  efficiently  with  one  of  these  alternatives  should  be 
effective  also  in  dealing  with  the  other. 

7.  Whether  due  to  true  overdominance  or  to  pseudo-overdominance,  the 
widespread  if  not  universal  occurrence  among  naturally  cross-fertilized  or- 
ganisms of  an  overdominance  type  of  response  to  inbreeding  and  outcrossing 
poses  a  problem  which  the  breeder  cannot  afford  to  disregard. 

8.  Neither  overdominance  nor  pseudo-overdominance  can  be  called  upon 
to  explain  the  differences  in  vigor  between  different  varieties  of  wheat,  beans, 
sorghums,  and  other  self-fertilized  forms.  Such  differences  are  determined  by 
genes  in  the  homozygous  state,  as  are  also  the  differences  between  homozy- 
gous inbred  lines  of  corn. 

ROLE  OF  LIMITING  FACTORS 

A  consideration  of  the  role  of  limiting  factors  in  quantitative  inheritance 
leads  us  to  a  third  group  of  postulates: 


GENE  INTERACTION  IN  HETEROSIS  325 

1.  The  adequacy  of  a  diet  is  determined  not  by  those  constituents  which 
are  present  in  ample  amounts,  but  by  those  which  are  deficient  to  the  point  of 
acting  as  limiting  factors.  Similarly  the  excellence  of  a  genotype  is  deter- 
mined not  by  its  stronge